note_output: don't send notes from a patch being stored

[dpwe]

Follow-up noted in #1191 and #1194.

The bug. patches_store_patch parses a patch string into the patch's own delta list. A segment like i1n60l1 names a synth and a note, so it goes down patches_event_has_voices, and note_output acted on it. As a result, storing a patch:

• sent a real note-on out of the MIDI port, or raised a CV gate, at the moment the patch was saved;
• if the patch string contained i1l0, ran ALL NOTES OFF, dropping a gate that a note being played was holding up.

Reproduction on main:

i1iv1in1
i1iG2,1           # synth 1: MIDI note output
K1024ui1n60l1     # store a patch  ->  MIDI OUT 90 60 127

The fix. note_output_handle_event(e, live), where live means queue == &amy_global.delta_queue, the same test #1194 uses for midi_cc_output. For a stored event:

• a note is still claimed, so a voiceless note-output synth skips the voice path exactly as it does when playing;
• nothing is sent, and the held-note stack isn't touched.

Tests. test_note_output.c gains test_storing_a_patch_sends_nothing, which covers:

• MIDI mode;
• CV mode, including that the next played note still raises its own gate edge;
• a stored panic leaving a held gate alone, while a played panic still drops it.

4 of its 6 checks fail on main and all pass with the fix. make ctest passes; make test is unchanged (90 / 43 at ~-99 dB, same as main here).

This and #1194 both touch patches.c, but in different places (yield_synth_commands there, patches_event_has_voices here), so either can merge first.

🤖 Generated with Claude Code

https://claude.ai/code/session_01PykXkeeWw2aRpC6PLfpqTQ

---

Generated by Claude Code

[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 @ 1e3b49d	this PR	Δ
1	993	997	+4
2	1143	1147	+4
3	1721	1728	+7
4	1877	1885	+8
5	2497	2504	+7
6	2626	2630	+4

Full chord settled render μs: 2627 (was 2619, Δ +0.3%) (peak 2632, 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#1380

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