I assume you took more pictures than the ones in your dropbox archive?
Regarding the windows, that’s what I would do, though I’m not the expert on architectural stuff (waiting for Götz here):
I wouldn’t mask them.
After reconstruction I’d probably cut the distorted reconstructed glass planes and add new 1-polygon planes in. Assuming they’re all on roughly the same axis, maybe 1 plane would be enough per floor.
Next up Retopo, then reproject texture and maybe paint the glass planes a single color, or add fake reflections.
Marco, what exactly do you want to know? Windows is always a challenge and most often it is not possible to get perfect results, if any at all. Even with Lukas’ clever suggestion, the texture in the window screens will be very bad because RC has contradicting info. In terms of geometry there should be a plane (which can sometimes be modeled if the windows are very dirty) but the reflections are different in each chot, so there will be a crazy mix. I sometimes just filled the panes with a uni color in post (in the orthophoto) but nowadays I just accept it as a limitation of the technique…
The rest of the building is as perfectly suitable as it can get, so there should be no problem.
As Luksa said, you need many more images for a succesful reconstruction though, I’d say 20 times as much at least and then you won’t get all the small details yet.