amyboard: report full sketch traceback (line numbers) to web editor

[bwhitman]

What

When an AMYboard sketch fails to load, the device now reports the full Python traceback (with file/line numbers) back to the web editor instead of just the one-line <ExcType>: <message>.

Why

The traceback was already being generated by sys.print_exception() — it just went to the serial console only and never reached the website. The 'X' sysex error frame carried only type(e).__name__ + ": " + str(e).

Note: simulate mode (run-in-browser) already shows the full traceback today, because the web scrapes sys.print_exception output off the JS console (spss.js:17). This gap only affected control mode on real hardware, where stderr goes to the serial port, not the browser — which is exactly the path a user on real hardware hits.

How

• New _format_exc_for_report(e) helper in tulip/shared/amyboard-py/amyboard.py captures sys.print_exception(e, buf) into an io.StringIO and returns the string.
• The three sketch-load error call sites now send _format_exc_for_report(e) instead of the one-liner.
• Single sysex frame, no chunking. Real MicroPython tracebacks are ~150–300 bytes; a single frame reliably carries ~768 raw bytes over Web MIDI (ZDUMP_STREAM_RAW_CHUNK in amy/src/transfer.c). The helper bounds output to 700 bytes, keeping the tail (MicroPython prints most-recent-call-last, so the exception line is last) and prepending a ...(traceback truncated) marker if it ever trims a pathologically deep stack.
• Falls back to the old type: message one-liner if traceback capture ever fails — best-effort, never masks the original error or breaks the self-heal.
• No web-side change: show_python_error already renders multi-line text in a monospace <pre> with white-space:pre-wrap.

Deliberately out of scope

The sketch.loop() runtime-error path is left serial-only. It runs every ~60ms, so reporting from there would spam sysex frames / re-pop the modal every tick. The existing feature (and this PR) covers the load path, which fires once.

Testing

• Truncation logic unit-tested in isolation (short tracebacks pass through whole; deep ones truncate from the front but preserve the exception line; stays under the 1024 base64-char single-frame limit).
• Needs hardware verification on the Pi HW-CI bench: flash AMYboard firmware, push a sketch with a deliberate error (e.g. NameError on a known line), and confirm the web editor modal shows the full traceback with correct line N (not line 0 — confirms MICROPY_ENABLE_SOURCE_LINE is compiled in).

🤖 Generated with Claude Code

[github-actions]

🔌 AMYboard PR preview

Editor + flasher: https://amyboard-pr-1059.vercel.app/editor/

This preview bundles this PR's firmware — its flasher only flashes this build (not the release). Rebuilt on every push; removed when the PR closes.

The hardware CI has been kicked off and should return within a few minutes, stand by!

[github-actions]

🌷 Tulip Web PR preview

Tulip Web: https://tulip-pr-1059.vercel.app/run/

Flash a Tulip to this build: on-device run tulip.upgrade(pr=1059).

Rebuilt on every push; removed when the PR closes.

[github-actions]

🎛️ HW CI (physical bench)

AMYboard (USB-MIDI + AMY zP → audio): ✅ PASS — flashed this PR’s firmware; all checks matched the references.

Tulip (TULIP4_R11; serial-REPL audio + WiFi screenshot): ✅ PASS — flashed this PR’s firmware; all checks matched the references.

⬇️ Artifacts: recordings · screenshot · serial logs · run logs

_{Self-hosted bench. Audio spectral-compared to ref/hwci_basic.wav + ref/tulip_basic.wav; Tulip screenshot pixel-compared to ref/tulip_screenshot.png. Both analog outs share one capture card, so the tests run sequentially.}