Table with white/black background not being fully captured

Hi there! First time `camelot` user, and trying to detect the full table. The upper part with white background is correctly detected, the bottom with black background (and white text) is not detected as part of the table.

I've tried different options for `flavor` and `process_background` but still no success. Anyone knows if it's possible to capture the full table given that the lower part is in the opposite background/text colour?

The report provides a screenshot but no PDF, file, test, or named entry point. First obtain a reproducible PDF and compare extraction with the reported flavor and process_background options, then trace the table-detection path for mixed background colors. Done means the full table is captured or the limitation is documented with a regression case.

Written by the indexing model from the issue text.