Enhancements to PDF text position assertions:
- Aggregate all regex mismatches and report together instead of failing
  on first error
- Add granularity attribute to compare positions at ancestor structure
  elements (e.g., Div, P) rather than individual text spans
- Include page number in position assertion failures
- Fix duplicate error reporting for ambiguous text matches

Test runner fixes:
- Fix --keep-outputs flag for book projects
- Fix book output path detection for multi-file vs single-file formats

Command-line tool that displays the PDF tag tree structure with marked
content IDs (MCIDs) and associated text. Useful for debugging PDF
accessibility structure and understanding how content is tagged.

Features:
- Search all pages for text patterns
- Highlight matching structure elements
- Show MCID mappings and bounding boxes

Remove version auto-detection from prepare-dist.ts and require explicit
version parameter. Updates performance-check and test-bundle workflows
to pass the version from version.txt.

Ignore .quarto cache directories which are populated at render time.
Also add gitignore entries for book test directories.

Simplify the logic for resolving user filter script paths. The previous
code had redundant checks and incorrect path resolution for filters
specified with scriptFile in _extension.yml.

Bundle fontawesome, octique, showybox, and theorion packages from
Typst preview repository. These are copied to .quarto cache at render
time, enabling offline builds without network access to typst.app.

… commit d39d268a9

git-subtree-dir: src/resources/extension-subtrees/orange-book
git-subtree-split: d39d268a93a8e14518b59a2be05893f0a8943aef

…rces/extension-subtrees/orange-book'