fix(nas): refuse deletes while library storage is unreachable; real file removal; stat cache

Three changes from the network-drive audit:

1. Delete-ordering fix. Deleting a media item cascades its annotations,
   so the delete must never proceed while the item's library storage is
   unreachable — the DB row died, the file survived as an orphan, and
   the next scan re-imported it bare (annotations gone forever).
   resolveMediaStorage stats the library's folders: reachable → delete
   proceeds and the file is removed by ABSOLUTE path (the old code
   passed a library-relative path to os.Remove, which never resolved —
   file deletion has been a silent no-op); every folder unreachable →
   409 with 'storage unreachable', row and annotations intact.

2. Latency: resolveLibrary stats every library folder on every SSR
   render; folder reachability is now cached with a 30s TTL.

3. README: Network Storage section documenting the outage hysteresis
   (mark → archive → auto-restore), the 90-day retention purge
   (ARCHIVE_RETENTION_DAYS, 0 = never), refused deletes while storage
   is down, and watch mode not firing on network mounts.

Integration tests: storage-down bulk delete refuses with the row
intact; real-file delete removes the file from disk.

README.md

@@ -127,6 +127,32 @@ The first user to register automatically becomes an admin.
 
 ---
 
+## 💾 Network Storage (NAS / NFS / SMB)
+
+Libraries on network mounts are supported for everyday use — serving,
+reading, progress, and annotation sync all work over the mount — with a
+few behaviors worth knowing:
+
+- **Outage hysteresis.** When the mount drops, files miss one scan
+  (marked missing, hidden) then a second (archived, still recoverable).
+  When the mount returns — by path or by identical content elsewhere —
+  everything is restored automatically: annotations, progress, and
+  covers included.
+- **Retention purge.** Archived items older than `ARCHIVE_RETENTION_DAYS`
+  (default **90**, `0` = never auto-purge) are deleted for good by
+  library scans, annotations included. A NAS offline longer than the
+  retention window loses those books' metadata on return.
+- **Deleting while the storage is down is refused.** Deleting a book
+  cascades its annotations, so deletes are rejected with a clear error
+  while the item's library storage is unreachable. Archive the item
+  instead, or retry when the mount is back.
+- **Watch mode does not fire on network mounts** (inotify doesn't work
+  over NFS/SMB). The periodic scan is the change detector.
+- **Latency** — folder reachability checks are cached ~30s; large
+  libraries on slow mounts will still feel slower than local disk.
+
+---
+
 ## 🛠 Tech Stack
 
 - **Backend**: Go 1.25+ with Echo framework