midi_out: skip USB gadget when no host is enumerated (fixes MIDI clock out over TRS without USB)

[bwhitman]

Problem

Reported on Discord: with tulip.external_midi_sync(send=True) on AMYboard, MIDI clock goes out over both USB and the TRS MIDI out port while USB is attached — but with USB disconnected (eurorack power only), the TRS out goes nearly silent and the board appears to hang; a downstream device "occasionally responds".

Root cause

midi_out() always tries the USB gadget branch when AMY_MIDI_IS_USB_GADGET is configured. TinyUSB's tud_midi_stream_write has no mount check — it just fills the class driver's tx FIFO, and write_flush silently fails when no host has enumerated the device. So with no USB host attached, the FIFO fills after ~16 messages and every subsequent write returns 0, which sends midi_out into its stall-recovery loop (mp_usbd_task() + vTaskDelay(1) × 1000) — ~1 second per byte — before falling through to the UART write.

MIDI clock out ticks come from sequencer_check_and_fill() → midi_clock_out_tick() in the render path, so at 24 PPQ this wedges the sequencer and reduces the TRS/DIN clock from 48 bytes/sec to ~1/sec at 120 bpm — exactly the observed "occasionally responds" behavior. The same stall hits any midi_out caller (e.g. Tulip's tulip.midi_out()) once USB is unplugged.

Fix

Guard the gadget branch with tud_ready() (enumerated and not suspended). Unpowered / charger-only / detached USB skips the gadget write instantly and the UART gets bytes at full rate; behavior with a real host attached is unchanged.

Testing

• make test: 117 tests pass
• tulipcc AMYboard firmware (idf.py -DMICROPY_BOARD=AMYBOARD build) builds clean against this branch

🤖 Generated with 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 @ c829ff2	this PR	Δ
1	1385	1383	-2
2	1548	1543	-5
3	2136	2134	-2
4	2328	2324	-4
5	2975	2968	-7
6	3169	3163	-6

Full chord settled render μs: 3164 (was 3170, Δ -0.2%) (peak 3173, 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#1215

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