feat: optional Hardcover wishlist sync with automated, library-aware downloading

[InfiniteAvenger]

Summary

Adds an opt-in automation pipeline that syncs a Hardcover reading shelf into
Shelfmark requests and (optionally) downloads them, skipping anything already in an
Audiobookshelf library. Everything is disabled by default — with all toggles off,
behavior is identical to today.

What this adds

A new Settings → Hardcover Sync tab with three independently-toggleable capabilities:

• Hardcover wishlist sync — pull a Hardcover shelf (e.g. "Want to Read") into
  Shelfmark as requests on a schedule (configurable in minutes or hours) or via a
  "Sync now" button. Synced requests carry full metadata and cover art.
• Automatic downloads — auto-approve and download synced requests, choosing the best
  release from a drag-to-reorder source-priority list. A strict title/author/
  audiobook-format (and seeder) match guard avoids grabbing the wrong file; anything
  without a confident match is left as a pending request for manual review.
• Library-aware de-duplication — check an Audiobookshelf library (ISBN + fuzzy
  title/author) and skip books already owned. Fails open (proceeds) if ABS is
  unreachable, so the pipeline never stalls.

Design

• Opt-in, no default behavior change — all features gated behind settings that
  default to off.
• Reuses existing systems — synced books flow through the normal request →
  fulfil_request → queue_release download path; there's no parallel pipeline.
• One behavior-preserving refactor — extracts the per-source release search out of
  the /api/releases route into a reusable helper so the auto-picker and the route
  share it.
• Conservative matching — biased toward leaving items pending rather than
  downloading the wrong file.
• In-process scheduler modeled on the existing download coordinator (daemon thread).

New modules

core/text_match.py, core/release_search.py, core/auto_download.py,
core/hardcover_sync.py, core/hardcover_scheduler.py, core/library_index.py,
config/hardcover_sync_settings.py; plus wiring in main.py (scheduler start + search
refactor) and config/settings.py (tab registration).

Testing

• ruff clean on all changed files.
• Existing tests/core suite (request/release/search/requests-service) passes.
• Verified end-to-end against a live Hardcover account and Audiobookshelf instance:
  wishlist → request (with cover) → strict-matched auto-download, plus owned-book skip
  and fail-open when ABS is down.

Scope note

I'm aware the README lists library integration, automation, and monitoring as explicit
non-goals, so this may simply be out of scope — totally understand if so. I built it for
my own setup and figured I'd offer it in case it's useful. It's entirely opt-in and off
by default, so it has no effect on the core tool for anyone who doesn't enable it. Happy
to keep maintaining it as a fork either way.