treat ALL instruments as wave-managed (like morph) -- make native wave engine go away.

[thenfour]

don't do this until we're ready to make our own playroutine (#74) because of the waveform slot allocation complexity

the reason: generated wave slots eat into the waveform pool.

• reserve waveforms 0-3 for the 4 channels
• we will keep our own waveform pool outside of waveform memory (that can hold many more waveforms)
• during playback all notes, all sfx will have their waves written in realtime to wave memory.

2 channels playing the same sfx is actually troublesome... we don't want to duplicate sfx, or force sfx into a pool system, but we can't have them crossing channels because there will always be a scenario that causes glitches.

---

better idea: use waveform slots a bit like a very simple voice allocator. Not fixed waveform -> channel mapping. Keep a table of waveformIndex -> [sfx, channel].

• during playback, find the best waveform index to use per sfx,chan, use it as your "target slot".