[BeetleB]

> is very different from “there is a proof over there so I will accept it as true and you should too” (i.e., how citations tend to be used most of the time, unfortunately).

I have seen exactly this. A (famous) professor at my university had a manuscript for a textbook that he had been working on for years, but had not yet published. A number of people wrote papers citing the (unpublished) book.

Of course, this is a bit of an outlier as the author had over the years given people drafts of his book.

Still, I would say that as long as the referee can access the web site and verify it, it should be allowed (and I'll still assert that it has been allowed on numerous occasions). May vary with discipline, though.