bug#45052: 27.1; GDB completion moves point

[Eli Zaretskii <eliz@gnu.org>]

> Cc: 45052@debbugs.gnu.org
> From: martin rudalics <rudalics@gmx.at>
> Date: Sat, 5 Dec 2020 16:21:15 +0100
> 
>  > Martin, do we have any way of preventing the original problem via
>  > display-buffer actions, per chance?  I couldn't find anything to that
>  > effect, FWIW.
> 
> There is `set-window-buffer-start-and-point' for this purpose but it is
> not used by `display-buffer'.

OK, thanks.

> I'm afraid that with these new completion frameworks proliferating
> we may have to check our sources for more incarnations of
> `set-window-start/-point' couplings that do not set NOFORCE.

Maybe we should teach these functions to set the optional_new_start
flag of the window, not only the force_start, given some special value
of NOFORCE.  That'd be the next thing I'd try if Philip (or someone
else) reports some adverse effects from the simplistic change I
proposed.