feat: add ja.mangarawbest

[kinboy56]

Fix #16

Summary

• Search, browse/filtering, series details, chapter list, and page list
• 3 listings: Popular, Weekly Popular, Newest
• Filters: sort (8 options), search-by name/artist, status, and genre (52 options, include/exclude)
• Selectable image server (site ships 3 CDN mirrors; some regions can't reach the default one)
• Deep linking for both series and chapter URLs

Implementation notes

• /search?q=... silently ignores the query — it always returns the unfiltered catalogue. The actual search endpoint is filter[name]=... / filter[artist]=... on /manga-list, reverse-engineered from the site's own app.js. filter[doujinshi]=... (the third "search type" option in the site's own UI) never returns any results regardless of value, so it was left out of filters.json rather than shipping a filter option that silently returns nothing.
• Genres appear twice on series pages: once in the actual info block, and again in an SEO keyword cloud at the bottom of the page that links to the same /genre/... URLs but prefixes every label with the series title (e.g. <title> Action). Tags are scraped from a selector scoped to the info block only, and there's a test asserting no tag contains the series title, so a regression here fails loudly instead of silently reintroducing junk tags.
• Chapter titles carry no information beyond the number — every chapter across the sampled catalogue is titled exactly 第N話, which the app already renders from chapter_number. Keeping the scraped title as well would show it twice in the reader (e.g. "41話 - 41話"), so titles that are just 第<number>話 are dropped.
• Dates are relative-only (6日前, 2週間前, etc.) — there's no absolute date anywhere in the markup, so they're converted from current_date().
• Status label text is inconsistent: the "completed" badge reads "完了" on some pages, and the filter link it sits inside consistently uses filter[status]=1. Status is parsed from that query value first, falling back to the label text only if the link is missing.
• The source has no explicit content rating of its own — content_rating is inferred from genre tags (成人向け/hentai-family tags → NSFW, Ecchi/Mature/etc. → Suggestive, otherwise Safe), since the source itself is registered as contains-nsfw.

Test plan

• cargo test — 19 tests, all hitting the live site (no mocked HTML), covering: browse pagination (including the single-page/no-results case where the pagination bar is absent), all 3 listings, search hits, filter combinations actually narrowing results (asserted against the unfiltered baseline, not just "doesn't error"), series details, chapter list ordering/format, page list, and deep link parsing (including www. / query-string / trailing-slash variants)
• cargo fmt --check — clean
• cargo clippy — no warnings
• aidoku package + aidoku verify — schema validation passes for source.json, filters.json, settings.json, and the icon (128×128, fully opaque)
• Manually installed via aidoku serve and read a chapter in the iOS app

Checklist

• The source can be compiled without any warnings or errors
• cargo fmt has been run before submission
• cargo clippy outputs no lint warnings
• All files have an additional newline at the end
• JSON files use tabs for indentation
• status / content_rating are set on series when available
• DeepLinkHandler is implemented for both series and chapter URLs
• Home page / listing components beyond simple sort-based listings — not implemented (optional per CONTRIBUTING.md)