Audible/audnex metadata can contain HTML entities such as `"` and
`&` that were passed through untouched, so titles like
`The "Hitler" Myth` were displayed literally instead of
`The "Hitler" Myth`.
Decode the entities at the ingestion boundary in `cleanResult` using the
existing `htmlSanitizer.stripAllTags` helper (the same plain-text
treatment already used for collection/playlist names). This decodes once,
avoiding the double-decoding of legitimate literal entities.
Fixes#5340
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
