> Because the section is called "VC Undo", I guess.

OK.  But I fail to see a reason why a user, after reading the Undo
section, would expect that M-x undo acts upon anything that is not
buffer text.  That is what made me think (perhaps wrongly) that the
sentence is about the old behavior of revert-buffer.

At the least, I think the documentation could be improved.  Some of the
things vc-revert does can be reversed (e.g., bring back the reverted
changes in the affected buffers) and other changes cannot (or might not,
I'm not sure of the right wording here).  If that is correct, then both the
current text and my proposed patch fail to make that clear.