refactor: drop AI/vision pipeline + plain Postgres + full-refresh script

Removes the OpenCLIP-on-ONNX classifier and everything that fed or
consumed it:
  - backend: app/services/vision/, app/tasks/vision.py,
    app/services/feature_flags.py, app/routers/features.py — all
    deleted; admin AI/feature-flag endpoints and the worker-vision
    bootstrap call gone. Photo.needs_review and its index dropped.
  - frontend: AI Settings tab, useFeaturesQuery hook, FeatureFlag
    types, "Needs Review" sidebar entry + filter, needs_review filter
    URL param all gone.
  - infra: worker-vision compose service + models_data volume deleted;
    worker-light command no longer runs bootstrap_models; the db
    image switches from pgvector/pgvector:pg16 to postgres:16; backend
    Dockerfile drops the dedicated torch RUN layer; requirements.txt
    drops torch/torchvision/open-clip-torch/onnxruntime.

Alembic 0019_drop_ai_remnants:
  - drops photos.needs_review + ix_photos_needs_review
  - DROP EXTENSION IF EXISTS vector (must run before the image swap;
    the new postgres:16 doesn't ship pgvector)

New scripts/full_refresh.py: one-shot DB ↔ filesystem reconciliation.
Runs cleanup_data_integrity, scans every active SourceRoot inline
(no celery dependency so the worker can be stopped), hard-prunes
photo + folder rows for files that are gone, removes orphan
/data/thumbs/{user}/{photo}/ directories. New helper
prune_orphan_thumbnails in cleanup.py.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

backend/app/tasks/celery.py

@@ -2,15 +2,12 @@
 Celery configuration and app initialization
 """
 import logging
-import os
 
 from celery import Celery
-from celery.signals import worker_process_init
 from app.config import settings
 
 logger = logging.getLogger(__name__)
 
-# Create Celery app
 celery_app = Celery(
     'mulita',
     broker=settings.celery_broker_url,
@@ -19,43 +16,31 @@ celery_app = Celery(
         'app.tasks.scan',
         'app.tasks.thumbs',
         'app.tasks.video',
-        'app.tasks.vision',
-        'app.services.metadata',   # extract_metadata lives here
+        'app.services.metadata',
     ]
 )
 
-# Configure Celery
 celery_app.conf.update(
     task_serializer='json',
     accept_content=['json'],
     result_serializer='json',
     timezone='UTC',
     enable_utc=True,
-    # Robust acknowledgment: keep message in broker until task succeeds.
     task_acks_late=True,
     task_reject_on_worker_lost=True,
-    # Global time limits — individual tasks can override via decorator.
-    task_soft_time_limit=300,   # 5 min — raises SoftTimeLimitExceeded
-    task_time_limit=600,        # 10 min — SIGKILL
-    # Explicit routes for every task name. Wildcard patterns don't match
-    # short names produced by @shared_task(name='...').
+    task_soft_time_limit=300,
+    task_time_limit=600,
     task_routes={
-        # Vision queue — CPU-bound binary classification
-        'classify_content': {'queue': 'vision'},
-        'vision_fanout': {'queue': 'vision'},
-        # High-priority queue — thumbnails & duplicates
         'generate_thumbnails': {'queue': 'high'},
         'regenerate_all_thumbnails': {'queue': 'high'},
         'backfill_phashes': {'queue': 'high'},
         'regroup_duplicates': {'queue': 'high'},
         'incremental_regroup_duplicates': {'queue': 'high'},
-        # Low-priority queue — scans
         'scan_folder': {'queue': 'low'},
         'scan_all_source_roots': {'queue': 'low'},
         'backfill_gps': {'queue': 'low'},
-        # Pre-transcode HEVC videos in the background so /playback is a
-        # cache hit on first user click. CPU-heavy but tolerant of the
-        # low-priority queue (it doesn't block any user-facing flow).
+        # CPU-heavy but tolerant of the low-priority queue (doesn't block
+        # any user-facing flow).
         'pretranscode_video': {'queue': 'low'},
         # `watch_folders` is retired (file events come from NC webhooks)
         # but the task definition still exists as a no-op shim for any
@@ -79,21 +64,3 @@ celery_app.conf.update(
         },
     },
 )
-
-
-@worker_process_init.connect
-def _warmup_vision_models(**kwargs):
-    """Pre-load vision models in the worker process so the first task
-    doesn't pay cold-start latency. Only runs on the vision queue."""
-    # The worker name contains the queue — only warm up vision workers.
-    worker_queues = os.environ.get("CELERY_QUEUES", "")
-    if "vision" not in worker_queues:
-        # Heuristic: check the celery command line for -Q vision
-        import sys
-        if "vision" not in " ".join(sys.argv):
-            return
-    try:
-        from app.services.vision.registry import registry
-        registry.warmup()
-    except Exception:
-        logger.exception("Vision model warmup failed")