There's a difference between creating something and then also promoting it. The FSF and the GNU project shouldn't encourage the creation of the plugin and they shouldn't promote it. However, they cannot stop anyone writing the plugin that exposes the AST in such a way to be usable by proprietary programs. The Emacs project also shouldn't promote this project.
This is exactly similar to how FSF and GNU don't promote the Windows version of Emacs. They don't hinder it or stop it, but they don't promote it or mention it either.
I think the Emacs maintainer blew things out of proportion and so has everyone else.
RMS would very likely respond to this by frequently changing the plugin interface and updating only the blessed plugins. Linux is not entirely dissimilar in how it does device drivers.
> However, they cannot stop anyone writing the plugin that exposes the AST in such a way to be usable by proprietary programs.
I'm not entire sure they couldn't... It really comes down to the definition of a derivative work. It's often assumed that a process boundary would somehow raise some legally impenetrable wall between works, but I'm not sure a court would necessarily agree.
This is exactly similar to how FSF and GNU don't promote the Windows version of Emacs. They don't hinder it or stop it, but they don't promote it or mention it either.
I think the Emacs maintainer blew things out of proportion and so has everyone else.