Extract Style

Extract a style_guide JSON from an uploaded PPTX/POTX.

Auth: any logged-in user (no rate limit yet — extraction is cheap).

Errors: 400 - non-pptx file or unparseable 413 - file too large 500 - python-pptx blew up on a malformed file we tried to parse

Returns: { "primary": "#RRGGBB", "accent": "#RRGGBB", "text": "#RRGGBB", "bg": "#RRGGBB", "secondary": "#RRGGBB", "fonts": {"primary": "..."}, "filename": "<original>.pptx", "size_bytes": int, "_evidence": { ... } # diagnostic counters }