Use generic regex for stripping Extension:Translate HTML markup
The HTML stripping only handled tags (for untranslated messages) but not tags (for outdated translations). When FuzzyBot marks translations as fuzzy after a source message update, it wraps them in these tags, which breaks JSON parsing.
Replace the hardcoded replaceAll() calls with a generic regex that strips all variants. Also replace $.Deferred with native Promise and use jQuery DOM creation for the error alert to avoid XSS risk.