Avoid overrunning osc table (#966); default num_voices to 1; let drums set their own oscs_per_voice.

[dpwe]

There was a bug in the code finding blocks of available oscs when allocating voices in patches_load_patch; code was reworked. Testing revealed that loading drum patches involved allocating twice, because the drum patches start by setting their own num_voices and oscs_per_voice. Made num_voices default to 1 if not specified (and if no existing value) in patches_load_patch. This simplifies loading drum patches to amy.send(synth=10, patch=384) etc. Updated documentation.

[github-actions]

🎛️ AMY HW CI (AMYboard bench)

Flashed this PR's AMY (LoadTestChord: 6-voice Juno patch=1, one held note every 2 s) onto the physical AMYboard and measured the smoothed render load as the chord grows — back-to-back with the same sketch built at the PR's merge base, so Δ is this PR's own cost.

✅ PASS — the bench ran the test to completion.

notes held	main @ 0e0508c	this PR	Δ
1	1379	1377	-2
2	1548	1546	-2
3	2132	2135	+3
4	2323	2318	-5
5	2963	2963	+0
6	3161	3159	-2

Full chord settled render μs: 3158 (was 3159, Δ -0.0%) (peak 3166, 39 samples)

⬇️ Artifacts: serial log · load trace · report

_{Self-hosted bench (amyboardci). FAIL means only that the test could not run — the load values are informational, with no threshold and no audio compare. See tools/arduino_loadsweep/.}

[bwhitman]

⛓️ tulipcc integration PR opened

This merge was pinned into tulipcc for full-system CI: shorepine/tulipcc#1230

Test it there and merge that PR to move tulipcc onto this AMY.