I think you'll be surprised about that. I'm not in the robotics field (though perhaps I should be) but at an earlier stage in life I spent several years on remodelling and building houses - as in swinging a hammer, not singing deeds and checks.
Repair/remodeling is challenging because you need (approximate) knowledge of construction techniques from across a 150 year timespan, buildings settle and change shape over time, and generally owners don't usually want the building completely gutted and rebuilt from scratch. But it's not that hard. New construction is relatively straightforward. The biggest challenges for construction robots will be handling dust/debris (25% of construction work is cleaning things up) and maintaining health and safety (of the robots themselves; self-preservation is a skill that has to be learned).
That said I don't think you'll see this with the current generation of humanoid robots. A lot of the design assumptions seem almost comically wrong to me.
Knowledge is irrelevant. Idk why you'd think dust would be a problem too, that's also comparatively trivial. Health and safety is also a good one, lol.
It all comes down to robots being unable to hold and operate anything in 3d space, much less precision tooling in arbitrary environments, which is basically what construction is. Everything in construction needs extensive planning, then to be done in the exact right order, with almost perfect precision.
The idea that you could tell a robot how you want a house remodeled and that it'd be able to go from there to actually do it is frankly asinine.
We don't have robots at all, much less humanoid ones, not even doing comparatively trivial, fixed tasks like laundry or the dishes, and you think we're going to have them manage the infinitely complex task of remodeling a house? Your mental model of robotics is completely wrong.
It's better than your reading comprehension.