Tablet pressurecurve calculation is too narrow and differs from X11

The tablet pressure curve that's applied via the xf86-input-wacom driver differs significantly to the one applied to by mutter. And I'm not sure the Wayland pressure curve has much of a noticeable effect at all (I'm not an artist, cc @jimmac for a comment). Visualised the difference is quite obvious. This is the Wacom driver:

And this is mutter's curve for the same sets of values:

So it looks like the algorithm we use in mutter is ... wrong? incomplete? This was implemented way back in 2016 in commit f9552bb9 but I can't tell if it was intentionally a different algorithm. @carlosg?

Source data

• mutter.gnuplot
• xf86-input-wacom PR#326 to have the pressurecurve code accessible via a helper tool. This tool spits out this output when used with gnome-control-center's hardcoded pressure curves: gnuplot.data
  • Visualization with driver.gnuplot