Added:
- Improved content type detection with a regression test.
- Updated magic.py to fix fallback for page type detection (plaintext, HTML, etc.), addressing an issue where not all `text/html` types were correctly caught (related to issue #4302).
- Fixed sanitization of lone Unicode surrogates before filtering JSON (issue #4293).

Changed:
- The timestamp for the HEAD entry updated from 2026-08-04 09:11:33 +0200 to 2026-08-15 18:59:59 +0200.

Removed:
- No removals reported.