[PATCH libinput 0/4] trackpoint: switch to a new acceleration method

[Peter Hutterer]

I've been working on this on and off for too long, it's time to get this
merged and adjusted later if need be.

Basic summary: we used to use the low-dpi acceleration method for
trackpoints, this patchset switches us to a basic linear acceleration
method. Many trackpoints have an acceleration function built-in already,
putting our own one on top means we mess around with the shape of the
acceleration curve with possibly unpredictable results (e.g. our plateau of
acceleration may differ).

This set adds a hwdb entry for the trackpoint's reasonable range which 
tells us which deltas we need to care about. Since the ranges I've seen are
between 15 and 90 for maximum deltas [1], we can't apply a sane function
without knowing something about the hardware.

We also ignore/undo the two systemd properties CONST_ACCEL and SENSITIVITY
so it doesn't matter anymore whether those are set or not. It was a good
theory to make these more generic but it's simply insufficient. CONST_ACCEL
never really was that in libinput anyway and no-one else used it (to my
knowledge). The sensitivity had some effect of course (it triggered a
hardware property), this set merely makes it obsolete so that the same
acceleration behaviour is felt regardless of the sensitivity set.

Cheers,
  Peter

[1] let's ignore that one device with a delta range of 1.