I've been looking through HotU, at least partly from a speedrun point of view. (Also I tried to work out how the ending works. My main conclusion is that it doesn't :-P)
One thing I noted is that AFAICT, the only familiar clip is in Chapter 1, right (and it's a nonstandard sort of clip that exploits an oversight in the scripting, rather than a "true" clip)? This implies that we might be able to route out the Wizard levels entirely. So I looked for the second-fastest way to solve HotU Chapter 1 Level 1. I'm pretty sure that solving the control rod puzzle is faster than the alchemist's fire route (because that route requires you to run to both ends of the map); unlike what some casual players might think, the puzzle can be solved with just two rods (most but not all combinations work), and there are multiple places you can find each rod. In other words, the level works a bit like chapter 2 of the original campaign, with a variety of possible solutions and any% being a very low completion percentage. You can get two rods by randomly teleporting around if you're (very) lucky (whenever you take a teleporter, it leads to a random different teleporter, rerolled each time), but unfortunately it's not a combination of two rods that can solve the puzzle. However, there's a room fairly near the entrance which contains a white rod and a yellow rod in adjacent chests (possibly this is a mistake by the developers?), and that is a combination that can solve the puzzle. To reach it, turn left at the pillars, follow the corridor round to the right, and then use the door on the left that goes to a different area. The second room of that area contains the chests. (I've actually made a list of all the rod locations, which I can post if anyone's interested; I'm pretty sure white+yellow is the fastest combination, though.)
Unfortunately, it's not quite trivial to pick up the rods with an "unusual" speed character; one of the chests is unlocked, but the other is locked and has hardness 40 (!). However, it also has only 20 hp (meaning that elemental damage can destroy it without too much effort, although slower than I'd like), and its lock is vulnerable to Knock or to Open Lock (DC 28; as there are Thieves Tools +10 in the Armory, and 20 (base value for skill use outside combat) + 10 (from the tools) + 1 (minimum Open Lock skill) - 2 (minimum Dexterity) is 29, any character with lock opening ability can pick the lock if they can avoid being in combat at the time (the room is, unfortunately, full of enemies). Incidentally, if the character has a lot of Open Lock skill (i.e. maxed), they can plausibly pick the lock even inside combat. Something I haven't checked is whether Sharwyn can be persuaded to pick the lock, but getting her to do that while the room's full of enemies seems fairly implausible.
Part of the reason I've been testing this is that, unlike in the original campaign, it seems reasonable that a lot of time could be saved at the start of HotU via the use of a less custom character; level 1 is free, but levels 2 to 15 are running on the timer. Being able to use autolevel would be nice, and being able to use a pregenerated level 15 even nicer, as long as there isn't too much of a cost incurred by the suboptimal character choice later on. The most obvious choice is Rebekah Stormcoast, who's a level 15 monk (and thus runs a little faster than a level 14 or 12 monk with a familiar would). I've confirmed that she's at least capable of clearing Chapter 1 by herself in what looks like a plausible speedroute (making a detour to pick up and use the rods, but otherwise using basically the same route as in the HotU WR Mejcel linked earlier in the thread); and she does it by brute force (opening the chest with sonic damage from gloves, and just tanking hits rather than being invisible). However, I didn't get anything like a good time (door clipping while enemies are attacking you are a pain).
More plausibly, it's probably possible to do something like this: build a monk 1; start a game of HotU; buy a Wizard level; export the character; restart a new game of HotU with the exported character, and choose to autolevel to 15. I'm not 100% sure, but I think that would buy 14 monk levels automatically, and configure based on package. There might be a monk package that has a good set of skills.
(Silly idea: start a game of HotU; level up the character, automatically or manually; export the character; then enter it in one of the other two campaigns, and cruise around with a character that's 14 levels overlevelled. I'm suddenly really curious as to whether this would comply with SDA rules.)
Meanwhile, I doubt Chapter 2 is significantly improvable (unless it turns out that there's a faster way to get a cutscene including the Valsharess). Sudden realisation: is it possible to kill the Chapter 1 Valsharess, and what happens if you do? (My guesses are "no" and "nothing", but I'm sufficiently non-confident in them that this has got to be worth testing; there's at least one vaguely plausible way to screw it up that would skip directly to chapter 3, which would at least make the experience route very interesting.) Interestingly, the glitched Chapter 2 Valsharess boss fight is one that never comes up in the intended gameplay (and in fact, it's significantly more difficult), so speedrunning has managed to create a rather fun and interesting challenge accidentally. (AFAICT, it's in the same room that the intended fight is in, involving the same Valsharess, but the room is changed from its cutscene layout to a 1-on-1 battle layout during the intended sequence of the game, whereas the glitched fight is inside the scene from the cutscene.) The other thing that I learned doing a HotU testrun is that the cutscene warp glitch is much more difficult than the video makes it look; I can do it, thus confirming to anyone who might potentially be sceptical that the glitch is a real one, but it took me way more tries than it should have done.
Chapter 3 does look somewhat improvable, though; there's a lot of running around in the HotU record video that doesn't seem like it should be strictly necessary. If it weren't for the gold grind, I'm fairly sure the fastest route would be to go straight to the temple, and then just outright kill the Sensei and take her Amulet. (With the right dialogue path, the Amulet's enough to unlock the path to the Puzzle Ring if you have sufficiently high Wisdom. Getting Wisdom high enough should be trivial given our access to epic-level equipment, if it doesn't make it manually; IIRC, the required value is fairly low.) There should be only a minimal need to mess around in the Cania hub area.
As for the gold grind itself, I'm not sure how necessary it is, but I'm also not sure if killing the final boss could be done at all quickly (apparently, the developers were aware of the normal speed kill trick, as many of the final boss's immunities seem to be targeted at the quick kills specifically; why else would it have effects as narrowly specific as Immunity: Movement speed decrease and Spell Immunity: Harm?). If we turn out to need the 700k gold to buy a sequence break, then I think the fastest way to do the later stages of the grind will be to buy as many of the item we're using as we can afford, then sell them all at once. (When testing out the gold grind, I used a highly enchanted plate armour which the blacksmith in the inn (who was also in Chapter 2, but has a higher gold limit now) sells in infinite quantities, and which produces quite a lot of gold per round; that means you don't have to go free Lazy at all. I don't know whether or not that's optimal.) I think I'd rather try to find a "quick kill" rather than needing to grind 700k gold, though; given the levels we're grinding to anyway, a kill is reasonable enough, but whether we can kill faster than the gold grind is another matter. Bonus points if it can be done only with Monk levels, or with Monk + a 1-level splash.
Incidentally, I also did a quick Diamond run of the OC in order to unlock everything it unlocks, and attempted to use my unpatched strategy. It didn't really go too well (because too much has changed), and in particular, I ended up with a completely nonsensical and unfocused character. The interesting thing is, though: I still won, despite the character making no sense (it was something like level 15 or 16 with the levels split between Wizard, Rogue and Barbarian in a completely suboptimal way). This makes me wonder if I'm currently overgrinding on unpatched OC (although one issue there is that even though the extra grind doesn't seem to be necessary to make things possible, it may nonetheless make them faster). It also makes me think about the idea of a minimum-experience run again; at this point, I suspect the biggest sticking point will be Maugrim (as most of the harder fights are skippable, but the requirement to kill Maugrim seems to be pretty ironclad; note that with all the ways to cheese the Morag fight, it can't really be considered difficult at this point).
One thing I noted is that AFAICT, the only familiar clip is in Chapter 1, right (and it's a nonstandard sort of clip that exploits an oversight in the scripting, rather than a "true" clip)? This implies that we might be able to route out the Wizard levels entirely. So I looked for the second-fastest way to solve HotU Chapter 1 Level 1. I'm pretty sure that solving the control rod puzzle is faster than the alchemist's fire route (because that route requires you to run to both ends of the map); unlike what some casual players might think, the puzzle can be solved with just two rods (most but not all combinations work), and there are multiple places you can find each rod. In other words, the level works a bit like chapter 2 of the original campaign, with a variety of possible solutions and any% being a very low completion percentage. You can get two rods by randomly teleporting around if you're (very) lucky (whenever you take a teleporter, it leads to a random different teleporter, rerolled each time), but unfortunately it's not a combination of two rods that can solve the puzzle. However, there's a room fairly near the entrance which contains a white rod and a yellow rod in adjacent chests (possibly this is a mistake by the developers?), and that is a combination that can solve the puzzle. To reach it, turn left at the pillars, follow the corridor round to the right, and then use the door on the left that goes to a different area. The second room of that area contains the chests. (I've actually made a list of all the rod locations, which I can post if anyone's interested; I'm pretty sure white+yellow is the fastest combination, though.)
Unfortunately, it's not quite trivial to pick up the rods with an "unusual" speed character; one of the chests is unlocked, but the other is locked and has hardness 40 (!). However, it also has only 20 hp (meaning that elemental damage can destroy it without too much effort, although slower than I'd like), and its lock is vulnerable to Knock or to Open Lock (DC 28; as there are Thieves Tools +10 in the Armory, and 20 (base value for skill use outside combat) + 10 (from the tools) + 1 (minimum Open Lock skill) - 2 (minimum Dexterity) is 29, any character with lock opening ability can pick the lock if they can avoid being in combat at the time (the room is, unfortunately, full of enemies). Incidentally, if the character has a lot of Open Lock skill (i.e. maxed), they can plausibly pick the lock even inside combat. Something I haven't checked is whether Sharwyn can be persuaded to pick the lock, but getting her to do that while the room's full of enemies seems fairly implausible.
Part of the reason I've been testing this is that, unlike in the original campaign, it seems reasonable that a lot of time could be saved at the start of HotU via the use of a less custom character; level 1 is free, but levels 2 to 15 are running on the timer. Being able to use autolevel would be nice, and being able to use a pregenerated level 15 even nicer, as long as there isn't too much of a cost incurred by the suboptimal character choice later on. The most obvious choice is Rebekah Stormcoast, who's a level 15 monk (and thus runs a little faster than a level 14 or 12 monk with a familiar would). I've confirmed that she's at least capable of clearing Chapter 1 by herself in what looks like a plausible speedroute (making a detour to pick up and use the rods, but otherwise using basically the same route as in the HotU WR Mejcel linked earlier in the thread); and she does it by brute force (opening the chest with sonic damage from gloves, and just tanking hits rather than being invisible). However, I didn't get anything like a good time (door clipping while enemies are attacking you are a pain).
More plausibly, it's probably possible to do something like this: build a monk 1; start a game of HotU; buy a Wizard level; export the character; restart a new game of HotU with the exported character, and choose to autolevel to 15. I'm not 100% sure, but I think that would buy 14 monk levels automatically, and configure based on package. There might be a monk package that has a good set of skills.
(Silly idea: start a game of HotU; level up the character, automatically or manually; export the character; then enter it in one of the other two campaigns, and cruise around with a character that's 14 levels overlevelled. I'm suddenly really curious as to whether this would comply with SDA rules.)
Meanwhile, I doubt Chapter 2 is significantly improvable (unless it turns out that there's a faster way to get a cutscene including the Valsharess). Sudden realisation: is it possible to kill the Chapter 1 Valsharess, and what happens if you do? (My guesses are "no" and "nothing", but I'm sufficiently non-confident in them that this has got to be worth testing; there's at least one vaguely plausible way to screw it up that would skip directly to chapter 3, which would at least make the experience route very interesting.) Interestingly, the glitched Chapter 2 Valsharess boss fight is one that never comes up in the intended gameplay (and in fact, it's significantly more difficult), so speedrunning has managed to create a rather fun and interesting challenge accidentally. (AFAICT, it's in the same room that the intended fight is in, involving the same Valsharess, but the room is changed from its cutscene layout to a 1-on-1 battle layout during the intended sequence of the game, whereas the glitched fight is inside the scene from the cutscene.) The other thing that I learned doing a HotU testrun is that the cutscene warp glitch is much more difficult than the video makes it look; I can do it, thus confirming to anyone who might potentially be sceptical that the glitch is a real one, but it took me way more tries than it should have done.
Chapter 3 does look somewhat improvable, though; there's a lot of running around in the HotU record video that doesn't seem like it should be strictly necessary. If it weren't for the gold grind, I'm fairly sure the fastest route would be to go straight to the temple, and then just outright kill the Sensei and take her Amulet. (With the right dialogue path, the Amulet's enough to unlock the path to the Puzzle Ring if you have sufficiently high Wisdom. Getting Wisdom high enough should be trivial given our access to epic-level equipment, if it doesn't make it manually; IIRC, the required value is fairly low.) There should be only a minimal need to mess around in the Cania hub area.
As for the gold grind itself, I'm not sure how necessary it is, but I'm also not sure if killing the final boss could be done at all quickly (apparently, the developers were aware of the normal speed kill trick, as many of the final boss's immunities seem to be targeted at the quick kills specifically; why else would it have effects as narrowly specific as Immunity: Movement speed decrease and Spell Immunity: Harm?). If we turn out to need the 700k gold to buy a sequence break, then I think the fastest way to do the later stages of the grind will be to buy as many of the item we're using as we can afford, then sell them all at once. (When testing out the gold grind, I used a highly enchanted plate armour which the blacksmith in the inn (who was also in Chapter 2, but has a higher gold limit now) sells in infinite quantities, and which produces quite a lot of gold per round; that means you don't have to go free Lazy at all. I don't know whether or not that's optimal.) I think I'd rather try to find a "quick kill" rather than needing to grind 700k gold, though; given the levels we're grinding to anyway, a kill is reasonable enough, but whether we can kill faster than the gold grind is another matter. Bonus points if it can be done only with Monk levels, or with Monk + a 1-level splash.
Incidentally, I also did a quick Diamond run of the OC in order to unlock everything it unlocks, and attempted to use my unpatched strategy. It didn't really go too well (because too much has changed), and in particular, I ended up with a completely nonsensical and unfocused character. The interesting thing is, though: I still won, despite the character making no sense (it was something like level 15 or 16 with the levels split between Wizard, Rogue and Barbarian in a completely suboptimal way). This makes me wonder if I'm currently overgrinding on unpatched OC (although one issue there is that even though the extra grind doesn't seem to be necessary to make things possible, it may nonetheless make them faster). It also makes me think about the idea of a minimum-experience run again; at this point, I suspect the biggest sticking point will be Maugrim (as most of the harder fights are skippable, but the requirement to kill Maugrim seems to be pretty ironclad; note that with all the ways to cheese the Morag fight, it can't really be considered difficult at this point).