Rework SSD1677 4-gray refresh into per-mode single-activation waveforms

[ainyan03]

Summary

Rework the SSD1677 4-gray refresh so that quality / text / fast dispatch to a shared absolute refresh path (Mode 2 face-parity recovery, parity-aware full-frame transfer, one Mode 1 activation) whose only per-mode difference is a static LUT, while fastest stays a differential monochrome update with a static LUT and software-computed transition groups.

• quality: oscillation prefix erases image history with a group-balanced net drive, preventing ghost accumulation over repeated refreshes; the tone-forming tail runs twice so the endpoints stay as stable as the vendor waveform ones
• text: white-first waveform, forms the four levels without a black flash
• fast: absolute 4-gray waveform derived from the community SDK, with the black darken return shortened to 28 frames (the excess overdrive rebounds toward gray during later undriven holds)
• fastest: 2-frame reverse-polarity prepulse plus 8-frame dose with state-aware holds outside the updated region; VCOM matches the panel offset so held pixels sit field-free
• absolute refreshes wait on the BUSY line without the 400 ms refresh-minimum floor
• the fastest fallback now records its display baseline, so fastest-only usage no longer degrades into a full refresh on every update

Two small library fixes ride along:

• Make QOI_DEBUG opt-in: lgfx_qoi.c unconditionally defined it, so error paths always called fprintf(stderr), pulling newlib stdio into every build; the png/jpg decoders reference no stdio, and this aligns qoi with them (define QOI_DEBUG externally to restore the output)
• Fix a -Wsign-compare warning in Touch_CST226::getTouchRaw(): the loop counter was int while count is uint_fast8_t, which warns on architectures where uint_fast8_t is unsigned int

Hardware validation

Waveform tuning and regression tests were run on an M5PaperMono SSD1677 panel. The quality / fast / fastest waveforms in this PR are the settled versions from that work (fast full refresh measures ~341 ms with the vendor-derived LUT; fastest keeps untouched black/gray/white pixels stable across repeated dirty-rect updates). The text waveform keeps the white-first structure; its group balance is inherently uneven, which is a known power-off ghosting trade-off of that structure.

Checks

• PlatformIO esp32s3 (pioarduino) build with the SSD1677 test firmware
• LUT group-balance decoder check (ssd1677_lut_balance.py) for the static LUTs
• git diff --check