Repository navigation
- PDF profiling screens page images before TOC extraction, and nests the PDF's TOCs in a tree. An appendix TOC is now grafted under its appendix instead of landing at the document root.
- A section may start on a TOC's last page, and each TOC node sits under the section open on its pages.
- TOC pages that list sections without page numbers are now kept as TOC pages. They get their own TOC node, so FINE skips them, but they are not anchored or calibrated, and the coarse sections are unchanged. Each TOC owns its own node (
TOC, orTOC p.<first page>when several share one parent). - When a PDF has bookmarks, the outline-vs-printed-TOC judge reads only TOCs with page numbers, so a PDF with only unnumbered TOCs keeps its outline.
Reviewed in #572. PR CI passed, including lint, type checks, tests, CodeQL, and secret scanning. No new migration.