Add MangaUpdates as the default manga and manhwa source

[AmrMsCLL]

Summary

MangaDex is a scanlation host, so licensed titles (especially Naver and Kakao webtoons) are often missing or have no English chapters, which leaves chapter counts and release checks blank. Its manga category also only excluded Korean titles, so Chinese manhua showed up under Manga.

• New MangaUpdatesService (public API v1, no key) for manga and manhwa, registered ahead of MangaDex so it is the default. MangaDex stays selectable, and existing MangaDex library entries keep resolving by their own source key.
• Browse: Popular is list_reading (most readers), Recent is this year's series by month1_pos (trending this month). Search uses MangaUpdates relevance.
• type is Manga or Manhwa per category, so manhua drops out of both.
• Adult genres (Adult, Hentai, Lolicon, Shotacon, Smut) are excluded server-side unless the reader opted in, and left out of the genre list.
• Details carry associated titles, story and art credits, origin status, latest_chapter and the volume count; release checks report latest_chapter.
• mangaupdates.com/series/<base36> links are recognized as manga or manhwa.
• Discover shows MangaUpdates sorts (Most read, Trending, Rating) and no status filter; MangaDex keeps its own. Changing the source drops genre, status and sort, since they are source-specific.
• Credits page and link placeholders name MangaUpdates, as its acceptable use policy asks.

MangaUpdates titles are often romanized (for example "Sousou no Frieren"); English names are in the alternate titles.

Tests

• New server spec for browse and search bodies, adult opt-in, rejected filters, details, releases, link recognition and the default-source order.
• New frontend spec for source-aware sorts and statuses and the filter reset on a source change.
• The registry tests pass the new connector; the DISABLED_SOURCES test disables both manga sources so it still proves a disabled category throws.
• npm run test, npm run lint and npm run build pass. The compiled connector was run against the live API for each category and section, search, details, releases, link recognition and genres.