Add debugging developer docs: AT32 recovery, bootlog, USB MSC, GCC preprocessing

[sensei-hacker]

Summary

Adds developer-facing debugging documentation that previously only existed as informal notes:

• docs/development/at32-flash-recovery.md — generic AT32F43x flash-protection recovery procedure (STM32CubeProgrammer, ST-Link/OpenOCD, serial bootloader), independent of any specific cause.
• docs/development/bootlog-debugging.md — the logHasOutput()/log-port-required gating pitfall, an I2C/OLED init diagnostic table, and raw-register-read technique. Fills a real gap versus the existing serial_printf_debugging.md, which only briefly mentions bootlog in passing; cross-linked both directions.
• docs/development/gcc-preprocessing-techniques.md — reducing/filtering arm-none-eabi-gcc -E output when comparing preprocessed target headers.
• docs/development/performance-debugging.md — PID-loop rate-limiting pattern (citing the real armingStatusDivider example already in fc_core.c), the tasks CLI command, and a backup-domain GPIO drive-strength caveat.
• docs/development/target-split-verification.md — multi-tool verification strategy (preprocessor diff + source-level dead-code check) for splitting one multi-board target directory into several.
• docs/development/usb-msc-debugging.md — composite CDC+MSC USB descriptor conflicts, with the real usbDevConfig()->type == COMPOSITE gating mechanism.

Testing

Documentation-only change (no .c/.h files touched). Every checkable factual claim was verified against the current source tree in this checkout:

• logIsEnabled()/logHasOutput() gating logic in src/main/common/log.c, and dashboardInit()/feature(FEATURE_DASHBOARD) gating in fc_init.c
• DEVFLAGS_USE_RAW_REGISTERS behavior traced through bus.c → bus_busdev_i2c.c → bus_i2c_hal.c
• updateArmingStatus() and its armingStatusDivider rate-limiting in fc_core.c
• usbDevConfig()->type == COMPOSITE in usbd_conf_stm32h7xx.c and the standalone-MSC fallback comment in usb_msc_h7xx.c
• All internal cross-links between the new/modified files resolve to real files in this PR

Two files from an earlier draft of this batch were dropped after verification: an AT32 EOPB0/SRAM investigation doc (the feature it investigates never landed in system_at32f43x.c — the branch was abandoned) and a bare debugging-scripts tips file (its one INAV-specific nugget was too narrow to justify a file on its own). An RP2350_PICO SWD debugging guide was also held back, since that target doesn't exist on maintenance-10.x yet — it'll land with that target's own PR instead.

Code Review

Reviewed with the inav-code-review agent, twice (once before, once after fixes). It caught and I fixed: a performance-debugging.md example that had updateArmingStatus() backwards (presented as an unaddressed rate-limiting opportunity when it's actually already rate-limited — corrected to cite the real pattern), and an OpenOCD command-namespace mismatch in the AT32 recovery doc (at32f4x unlock vs. the stm32f4x.cfg-based config actually shown). It also confirmed the batch is clean of vendor-datasheet reproduction and session-investigation narrative framing.

[qodo-code-review]

Qodo reviews are paused for this user.

Troubleshooting steps vary by plan Learn more →

On a Teams plan?
Reviews resume once this user has a paid seat and their Git account is linked in Qodo.
Link Git account →

Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center?
These require an Enterprise plan - Contact us
Contact us →