If someone withdraws their contribution, then:
* If the maintainers don't take down all tarballs of previous versions containing any of that code, they are still redistributing the withdrawn code, with a GPL2+ notice attached.
* If the maintainers don't either take down the git repo, or painstakingly reconstruct history without that person and then force push, they are still redistributing the withdrawn code, with a GPL2+ notice attached. (Reverting wouldn't be enough, it's still in the history.)
Either of these could be pretty disastrous if taken to the extreme, but it's obvious the withdrawn code would no longer be in the current release. Have RMS or ESR (or indeed Linus) made any comment on what the right to withdraw means in the context of a publicly accessible git repo?