bug#68213: 30.0.50; completion-preview-tests failure in --without-x build

[Stefan Monnier via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>]

>> Any chance you could try and help me figure out answers to those questions?
>>
>> At the very least we have a documentation problem because, as a user
>> reading the docstring of `mouse-wheel-down-alternate-event`, I have no
>> idea what this means or does and how it differs from
>> `mouse-wheel-down-event`.
>>
>> `mouse-wheel-down-event` is a defcustom because X11 failed
>> to define events for scroll wheel, and depending on your mouse, those
>> were not always mapped to the same `mouse-N` events.  That doesn't seem
>> to apply to `mouse-wheel-down-alternate-event`.
>
> mouse-wheel-*-alternate-event were initially meant to reflect a second
> type of event generated by the window system in response to scroll wheel
> motion.  At the time, it was set on XInput2 builds and no more, where
> wheel-* events are generated in addition to the customary mouse-*
> events.
>
> This arrangement was later modified before the release to set
> mouse-wheel-*-alternate-event to mouse-* wherever it had previously been
> nil, to provide for xterm-mouse-mode generating such events whatever the
> default value of mouse-wheel-*-event.
>
> As such, I suppose the difference is that between the mouse event
> generally emitted by the window system, and events which might be
> encountered under special circumstances.

You're describing the difference in terms of what they're currently set
to, but I'm asking the difference in their effect.  I.e. what's the
difference between

    (setq mouse-wheel-*-alternate-event A)
    (setq mouse-wheel-*-event B)

and

    (setq mouse-wheel-*-alternate-event B)
    (setq mouse-wheel-*-event A)

My patch is based on the idea that the difference doesn't matter and
that one of the two can always hold `wheel-up/down` so we can
hardcode it.

>> Do you happen to remember the bug numbers of bug titles or something to
>> help me find info about those obscure problems?
> I think searching the bug tracker for "xterm-mouse-mode" and "mwheel"
> will help, and I also recall that the PGTK port was involved in some
> manner.

Thanks.  I see that bug#50321 suggests we keep `mouse-4/5` in the
`mouse-wheel-*-event` for any build that can use `xterm-mouse-mode`.
It should be easy to adjust my patch to support that.
Tho bug#49803 suggests we could fix it in `xt-mouse.el` as well and make
`xt-mouse.el` emit `wheel-up/down` events instead.


        Stefa