…avIds

Delete HandleIDMetadataLinks and id_linker.go; remove UpdateIDSymlink and
RemoveIDSymlink methods from MetadataService; remove the .id sidecar write
block from WriteFileMetadata (read path preserved for Phase 2 compatibility);
remove FilterExistingNzbdavIds from QueueRepository, Repository, and the
BatchQueueAdder interface; remove related call sites in nzbfilesystem MOVE
handler and nzbdav scanner processBatch.

- Add MigrationRecorder interface to scanner package with UpsertMigration
  and IsMigrationCompleted methods
- Update NzbDavImporter to accept MigrationRecorder as second constructor
  parameter alongside BatchQueueAdder
- processBatch now checks IsMigrationCompleted before enqueueing (skip
  already-imported/symlinks_migrated items) and calls UpsertMigration for
  new items, then strips nzbdav_id from the queue item metadata leaving
  only extracted_files if present
- createNzbFileAndPrepareItem sets SkipArrNotification=true on all items
- batchQueueAdapterForImporter gains migrationRepo field and implements
  MigrationRecorder via ImportMigrationRepository.Upsert/LookupByExternalID

…lure

Wire s.database.MigrationRepo into handleProcessingSuccess and
handleProcessingFailure so that import_migrations rows are marked
imported/failed when the corresponding queue item finishes. Both
calls are non-fatal: failures are logged as warnings and do not
affect the main processing outcome. If no matching migration row
exists (non-nzbdav import), the UPDATE simply affects 0 rows.

Adds RewriteLibrarySymlinks in internal/importer/migration to walk a
library directory and atomically rewrite arr symlinks that point at
<nzbdav_mount>/.ids/<guid> to the final altmount path.  Introduces
DBSymlinkLookup adapter in internal/database, a new
POST /import/nzbdav/migrate-symlinks handler, and enriches the existing
status endpoint with migration_stats when available.

Phase 1: import_migrations DB table tracks every nzbdav import by GUID,
surviving queue resets. Scanner uses it for dedup (replaces
FilterExistingNzbdavIds), sets SkipArrNotification=true, and strips
nzbdav_id from queue metadata. Service marks rows imported/failed on
queue item completion.

Phase 2: POST /import/nzbdav/migrate-symlinks walks an arr library,
detects symlinks targeting <nzbdav_mount>/.ids/<guid>, looks up the
AltMount final_path in import_migrations, and atomically rewrites them.
BackfillFromImportQueue seeds the table from historical queue rows so
already-migrated users can skip directly to Phase 2.

Cleanup: id_linker.go deleted, HandleIDMetadataLinks call removed, all
.ids/ write paths removed from metadata service (read path preserved for
backward compat), FilterExistingNzbdavIds dropped.

Frontend: Phase 2 card with dry-run preview and Apply button appears
after Phase 1 completes or when migration_stats.imported > 0.

Users who already migrated via the old nzbdav flow can now access
the symlink rewrite tool directly from a top-level 'Migrate Symlinks'
tab without needing to run Phase 1 first.

… add Clear All Migrations

- Phase 2 symlink rewrite inside the NZBDav import flow is now gated on
  an empty queue (total_queued + total_processing == 0) with a live
  "Waiting for queue to finish" banner so users don't rewrite symlinks
  while imports are still in flight.
- Split the NZBDav section into nested tabs (Import / Migrate Symlinks).
  The standalone tab runs against already-imported rows for users who
  imported previously and only need to rewrite the arr library.
- Extract the symlink form into a shared SymlinkMigrationForm component
  reused by both entry points.
- Add DELETE /api/import/nzbdav/migrations + UI button to wipe all
  nzbdav migration rows, for the case where the user deleted imported
  files from AltMount and wants to force a full re-import.