This restructures library scanning into a set of dedicated components with narrower responsibilities.

The scan pipeline is now split into:

• LibraryFileEnumerator for file discovery and directory traversal
• LibraryScanSession for coordinating a scan request
• LibraryScanState for scan bookkeeping and progress state
• LibraryTrackResolver for reading files, playlists, archives, and cue-based tracks
• LibraryScanWriter for batched database writes
• LibraryMonitor for filesystem watcher ownership and monitoring

Directory traversal should also be faster and less blocking on network drives, since the new iteration path does less recursive directory-list construction up front.

This PR also improves scan UX:

• Library scans can be cancelled from the Library menu
• Active library scans can be cancelled individually from the library settings page
• The status bar now shows more detailed scan progress
• The status bar includes a cancel button beside the active scan progress