filter-touchpad: normalize for dpi on the touchpad-specific flat profile

Closes #591 (closed), closes #707 (closed), closes #777 (closed)

We discussed this change in the first of the issues mentioned above already, but didn't follow through with it yet for some reason.

Apart from the added normalize_for_dpi call, this also brings variable names in line with a nearby very similar function (accelerator_filter_noop_touchpad_flat, which is used for button-scrolling, two-finger-scrolling, etc). That function remains slightly different by not applying the constant acceleration factor (on purpose), so we can't just deduplicate them.

Interestingly, the function for button-scrolling acceleration already applied dpi normalization ... and even more interestingly, that originates in the flat profile for mice, where we established that the lack of dpi normalization in the function for cursor motion is intentional. Maybe we should fix that internal inconsistency too (by removing all traces of normalization over there) ... thoughts?