test: add hardware-validated smoke tests (xiao_nrf54l15, xiao_mg24)

[cumin777]

What

Adds per-board MicroPython hardware smoke tests, placed in example/ alongside the existing per-board test scripts (xiao_stm32c5_full_test.py, xiao_nrf54lm20b_full_test.py, ...):

• example/xiao_nrf54l15_smoke_test.py — 6/6 PASS on hardware
• example/xiao_mg24_smoke_test.py — 7/7 PASS on hardware

Both were executed against the v1.27.0 master builds on real hardware (2026-09-29).

RA4M1 is not included: no board on hand, hardware run deferred — see the status comment below.

What the hardware run corrected

1. mg24 user C modules are uppercase (ADC, RTC, CAN, PDM, LowPWR via MP_REGISTER_MODULE(MP_QSTR_*)), with the callable class inside the module (ADC.ADC, RTC.RTC). Consistent with example/boards/xiao.py and every shipped example; an early draft of this test did import adc / import rtc and "failed" — that was a test bug, not a firmware bug. The test exercises ADC.ADC(("adc0", ch)).read_uv() across channels and an RTC.RTC() datetime round-trip (the on-hardware validation of the ci(mg24): drop dead-fork PR override, apply repo v1.27.0 patches for USER_C_MODULES #38 USER_C_MODULES wiring).

2. nrf54l15 gc threshold 50K was unreachable: the zephyr port default CONFIG_MICROPY_HEAP_SIZE is 49152, leaving exactly 48000 usable bytes after GC bookkeeping (measured free+alloc == 48000). Threshold is now 32K.

How to run

mpremote connect COM<x> run example\xiao_nrf54l15_smoke_test.py

Both boards carry an onboard debug bridge (CMSIS-DAP + CDC UART) — the REPL lives on the bridge CDC port; the MCU's own USB peripheral is not wired to the USB-C connector, so no other USB serial device is expected. mpremote specifically: its raw-paste flow control is needed to push the ~5 KB script through the bridge. Final LED check is visual.

[cumin777]

Status before merge (2026-09-30)

Validated on hardware (2026-09-29): xiao_nrf54l15 6/6 PASS, xiao_mg24 7/7 PASS.

NOT validated — no boards on hand, recorded here so the gap is visible:

• xiao_ra4m1 — hardware run deferred indefinitely (no board available). Its smoke test is therefore intentionally not in this PR. Expectation per the build config: the renesas-ra port uses MicroPython'''s own TinyUSB (MICROPY_HW_ENABLE_USBDEV=1), independent of Zephyr Kconfig, so it should enumerate a USB CDC REPL — but that is unverified, and getting a REPL at all is precisely the on-hardware validation of the VDCEN patch (0002-renesas-ra-usb-ldo-vdcen.patch, carried from upstream micropython#16409 via ci(ra4m1): drop upstream PR #16409 checkout, carry USB LDO VDCEN as patch 0002 #39).
• xiao_nrf52840 — known-bad: firmware boots but never enumerates a serial port (fix(nrf52840): enable USB CDC ACM console — firmware was headless #40; CI green, real hardware silent — runtime-death analysis needs SWD, archived).

Docs gap for users: the wiki has no MicroPython usage page at all for nrf52840 and ra4m1. For nrf54l15/mg24 a wiki page exists but never mentions that the REPL arrives through the onboard debug bridge'''s CDC port (the MCU USB is not wired to the USB-C connector) — users without that context hit "where is the REPL" (see #12).