discover v0.1: scrape + search + import

- requirements: add recipe-scrapers 15.6.0
- mealie.import_from_url(): POST /api/recipes/create/url returns slug
- db helpers: insert_discovered_recipe, update_discovered_meta,
  set_discovered_status, list_discovered_recipes (FULLTEXT + JSON
  filters), count_discovered_by_status, get_discovered_recipe;
  discover-job CRUD + anti-zombie finalize + stuck-job recovery
- discover_recipes.py: daemon-thread runner (mirrors enrich pattern)
  walks a URL list; scrape_me → reshape to mealie shape → INSERT IGNORE
  → forge.enrich_recipe → flip raw → enriched. SEED_URLS curated
  starter packs for allrecipes / bbc / smitten / pinch / hbh.
- endpoints: GET /discover, GET /api/discover/search (q + cuisine +
  complexity + protein + meal_type + kid-fit + max_minutes + status),
  POST /api/discover/import/<id>, /reject/<id>, /scrape-start (seed
  or urls list), /scrape-status, /scrape-cancel/<id>
- discover.html: filter row + card grid + collapsible scrape panel
  with seed chips and url textarea + live progress poll
- nav: 'discover' tab on /, link card on /me
- boot recovery: fail_stuck_discover_jobs at startup

requirements.txt

@@ -5,3 +5,4 @@ Authlib==1.3.2
 PyMySQL==1.1.1
 cryptography==43.0.3
 rapidfuzz==3.10.1
+recipe-scrapers==15.6.0