I think those are fairly accurate groupings.
for better or worse, I’ve been grouping these by interface, because the affordances each has (or doesn’t have) plays directly into how it would deal with audio.
there’s the normal mixer interface stuff (faders, knobs) which is some vague form of a minimal, small footprint mono or stereo mixer (with the nearness concept sitting in between those two options).
then there’s gestural interface stuff (joystick, fsr/capacitive strips, trill, etc.), which has kinda centered around gestural panning/levels, with a focus on quad channels stuff or other instances where there’s simply more room for those gestures to be meaningful.
not sure if expandable is the right way to think about the last, but they generally seem to be digital devices that are configurable but feature little to no interface beyond patching audio into and out of them (maybe calling them headless makes sense). they could run scripts themselves, interface with other devices (norns), or be designed to live within a eurorack system, or any manner of other things.
to me, these feel like hard splits. attempting to blend these directions into each other in any significant way will likely result in confusion (for both building and using the thing).
