Thinking about that, I believe it's because the thing that vc-next-action
does depends on the status of the files in the fileset. So the statuses
have to be compatible.
The core concept of the proposed new command ‘C-x v V’ and ‘V’ in VC-Dired
is to deduce the fileset from the statuses of repository files, without
explicitly marking the files in VC-Dired and without using only the
current file as ‘C-x v v’ does.
Some dwim logic could be employed such as collecting only edited files
to the fileset, but I'm not sure if this will cover 100% of user needs.
Maybe some customizable variable could be added with a list of statuses
to specify what files to include to the fileset.