bug#48042: 26.3; Macros don't work with french-postfix input method

[Stefan Monnier]

> If you are asking how I'd like to proceed with this bug, then, as I
> wrote in another message, I'd like us to have a full understanding of
> why binding inhibit--record-char didn't work well enough in the case
> in point.  It did solve some of the problem we had before that change,
> but didn't solve all of it -- why?

BTW, another possible way to attack the problem is to arrange for the
code that pushes to `unread-command-events` to use events of the form
`(t . EVENT)` so as to tell explicitly that we don't want them recorded.

I haven't re-read enough of the thread and corresponding code to know
which approach might be more applicable here; I'm just mentioning it in
case someone had forgotten about that option (been there, done that ;-)


        Stefan