bug#62207: 29.0.60; Trying to remove non-existent key binding instead adds a binding

[Stefan Monnier]

> If this part is still controversial, I'm okay with only fixing the bug
> itself.  As for REMOVE, let's see what Stefan has to say in defense of
> his request, and take it from there.

I can't remember off-hand all the subtleties that can show up in corner
cases, but I'll simply note that the current removal code is already the
result of my pointing out several not-so-corner cases that the original
coder missed (even though he was not a beginner at the keymap game), and
this very bug report illustrates that the code is still buggy.

[ Which reminds me that we need a regression test for this.  ]

I think the removal code is "good enough" for uses in a user's init
code, but I'd rather add some recommendation to avoid on it in
ELisp packages.


        Stefan