fix(cache): harden ohli24 cached list parsing

- reject cached list/search html when rows lack parseable link and title
- skip malformed rows instead of failing whole list parsing
- record the cache parsing fix in the dated history log
- bump anime_downloader plugin version to 0.7.20

Co-Authored-By: First Fluke <our.first.fluke@gmail.com>

2026-03-23.history.md

@@ -0,0 +1,15 @@
+# 2026-03-23 Work History
+
+## Summary
+- Strengthened Ohli24 browse-cache validation so list/search cache only survives when at least one row contains both a link and a title.
+- Hardened Ohli24 list, auto-list, and search parsing against nested `post-title` markup and missing image attributes.
+- Prevented malformed rows from crashing the entire list response by skipping incomplete entries with warning logs.
+
+## Implementation Notes
+- Added `_extract_text()` and `_extract_first()` helpers in [`mod_ohli24.py`](/Volumes/WD/Users/Work/python/ff_dev_plugins/anime_downloader/mod_ohli24.py) to avoid repeated unsafe XPath `[0]` access.
+- Extended `_is_valid_cached_html()` to parse list/search rows and reject cache payloads that only contain placeholder rows.
+- Updated changelog and plugin version to `0.7.20`.
+
+## Verification
+- `python3 -m py_compile /Volumes/WD/Users/Work/python/ff_dev_plugins/anime_downloader/mod_ohli24.py`
+- Smoke-checked validator behavior for valid list/detail HTML and blocked HTML patterns during implementation.