As someone who picked Option 3 (but, as noted, who's good with the design direction generally commented on thus far), I'd like to come back to the loot question for a moment.
I got done with playing Two Worlds II a while back. I'd really been looking forward to it as a sequel to the original Two Worlds, which turned out to be a rather nice open-world game with some clever gameplay elements.
Unfortunately the sequel didn't feel as much fun to me. I won't belabor all the reasons; the point here is that as I was playing TW2 I found that I would get to the end of some dungeon, having obliterated all the enemies, only to discover that the reward was often an object I already had or, in numerous cases, nothing at all!
I'm not a highly loot-driven gamer. I enjoy seeing the sights, and can enjoy the journey as much as (or more than) the destination. But to hand-build some dungeon-based mission, to set up a bunch of progressively more difficult challenges, and then at the end provide the player with no interesting reward at all... even I found that not really fair, and less than fun. It felt like an unspoken agreement had been breached: "If you'll wade through all these puzzles and enemies, I'll make it worth your while at the end." (XP is usually not enough unless it happens to ding you to the next level.)
Which brings me back to A Valley Without Wind. Even if AVWW is not designed to be a highly loot-centric game, there's still that unstated promise that a game developer makes to potential players that the quality of the rewards will be proportional to the difficulty of the challenges. So how does the question of building sizes bear on that (or maybe the other way around)?
Small buildings can't be expected to provide much challenge; they're probably best treated as brief breaks from overworld and major dungeon exploration. Loot can be random stuff with the occasional nice bit (if you search for it) and most folks should be OK with that.
"Blocked" buildings could probably be treated relatively simply: to the extent that there's loot at all, the best stuff -- guarded by the toughest challenges -- should usually be found (or generated) near the point that's farthest from the entrance(s). If you made it there, it's not unreasonable to expect some relatively shiny bauble. Not a major game-changer, but a few goodies whose total value will help the player deal with new challenges.
Full-sized buildings, though... what kind of reward is proportional to the labor of clearing out the Augean stables of a 1000-room complex? For those souls who choose to take on such a challenge and actually complete it, should there be some special reward (especially if these big buildings are extremely rare)? Or is the total amount of "stuff" one may collect in such a building (including XP or its equivalent) sufficient?
To put it another way: if there's not some great prize waiting somewhere inside a 1500-room tower, are such buildings even worth implementing? I'm not rhetorically suggesting either "yes" or "no"; I'm interested in hearing what others think about it.
(Side note: speaking of loot, the original Two Worlds had a couple of really interesting features that the sequel didn't effectively follow up on. First, individual weapons and pieces of armor were numbered by type, and you could combine similar types to create objects with better stats. Secondly, armor came in "sets": if you could find and put on the Type 334 helm, chestplate, leggings, boots, and gloves, not only did it look nice, you got some useful special effect. The combination of these two features meant that even I wound up getting interested in loot as a kind of "gotta catch 'em all!" collection sub-game. In fact, the pull of this was incredibly powerful. I have no idea whether something like this would feel like a good fit for AVWW; I mention it because no mention of loot and Two Worlds would be complete without praising the effectiveness of these features at encouraging continued gameplay.)