Fix: log string deduplication - #2097

Detailed description

• No significant new features.
• The existing problem is suboptimal/repeating message strings for DEBUG_ERROR/WARN/TARGET().
• This PR solves it by factoring out unique substrings and enabling linker relaxation across newly-identical strings.

Discovered by staring at strings -n16 build-f411ce/blackmagic_blackpill_f411ce_firmware.bin | less output.

1. First commit saves up to 256 bytes when all targets are enabled and DEBUG_TARGET() is enabled.
2. In stm32h7.c, I drop the logline I added previously for H7B3 compat FLASH_CR patcher helper, because the driver is tested to work and that logline spams output during flashing. -116 bytes.
3. Renesas RA driver has three really long warnings with identical prefix/suffix and missing comma. -136 bytes.
4. LPC17xx & LPC40xx have similar mass-erase functions in which strings differ and hence don't get merged. -104 bytes.
5. LPC55xx reports IAP errors for five unique commands, and I decided to extract these instead. -52 bytes.

I would need some autobuilder script to measure size savings from each commit, but they're on the order of 50-100 bytes, for a total of about -292 bytes up to DEBUG_WARN, and -372 bytes more up to DEBUG_TARGET (-664).
Can't test on most of targets modified in PR scope. Idea similar to #1527 and #1863. Should have no effect on release firmware with disabled logging. When -Ddebug_output=true, default native and swlink/stlink/bluepill and f072 builds still overflow by 17-23 KiB. The other probe platforms are not 128 KiB but 256 KiB or even larger.

Your checklist for this pull request

• I've read the Code of Conduct
• I've read the guidelines for contributing to this repository
• It builds for hardware native (see Building the firmware)
• It builds as BMDA (see Building the BMDA)
• I've tested it to the best of my ability
• My commit messages provide a useful short description of what the commits do

Closing issues