libinput has a new trackpoint acceleration

[Peter Hutterer]

Just a heads-up, I just merged a branch that fixes trackpoint acceleration
in libinput. The previous approach was simply broken, the new one is quite
similar to what we had before anyway - calculating speed from the deltas and
applying the acceleration curve from that. The curve is adjusted for
trackpoints with a relatively wide configurable range.

Trackpoint data is mostly garbage and device-specific, so there's a magic
multiplier that is applied to the deltas. This multiplier replaces the
previous trackpoint range. Some more details are here:
https://wayland.freedesktop.org/libinput/doc/latest/trackpoints.html

I wish I could say this has been extensively tested by multiple users but
I've given up hope on that. Pointer acceleration testing feedback is usually
zero or so close to zero that the difference doesn't matter. It's been
tested on the machines I have access to and it works better there.

Any pointer acceleration bugs that aren't accompanied by either some patches
or workable algorithms will be closed as cantfix from now on.

Cheers,
   Peter