feat(b3): ml-worker becomes optional — embed-only role, decoupled GPU coordination, cpu-embed switch

The ml-worker's ONLY processing role is now the CPU whole-image embed fallback
(tag_and_embed renamed embed_image — Camie tagging was retired #1189 and the
name kept implying otherwise; videos were already handled agent-style: frame
sampling + mean-pool). Detection/cropping/CCIP stay GPU-agent-only, and their
completion is judged per-pipeline: ccip by gpu_job rows, siglip by concept
regions at the current model version — never by image_record.siglip_embedding.
A CPU embed therefore can NEVER close crop work for the agent (regression test
pins this; only the whole-image 'embed' job, the same artifact, is satisfied).

Making removal actually safe (operator will drop the container):
- GPU-queue coordination (enqueue_gpu_backfill, recover_orphaned_gpu_jobs,
  reprocess_gpu_jobs) moved verbatim to tasks/gpu_queue.py on the maintenance
  quick lane — it lived on the 'ml' queue only by module colocation, which made
  the ml-worker a hard dependency of the whole agent pipeline.
- New ml_settings.cpu_embed_enabled (migration 0074, default ON so agent-less
  installs keep working): OFF stops the four import hooks queueing embed work
  nothing will consume and no-ops the manual backfill; switch lives on the
  renamed 'CPU embedding backfill' card.
- NB heads training / auto-apply still run on the ml image (sklearn) — a stack
  that removes the container gives those up too.

Deploy note: in-flight messages under the old task names are dropped by the
new workers; the 60s orphan sweep + hourly backfill re-fire under the new
names immediately.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM

tests/test_tasks_ml.py

@@ -1,4 +1,4 @@
-"""tag_and_embed (embed-only) / backfill task tests. The pure _is_video helper
+"""embed_image (embed-only) / backfill task tests. The pure _is_video helper
 is a unit test; the DB-touching backfill query is an integration test with
 monkeypatched dispatch."""
 
@@ -23,7 +23,7 @@ async def test_backfill_enqueues_missing(db, monkeypatch):
 
     calls = []
     monkeypatch.setattr(
-        ml_tasks.tag_and_embed, "delay", lambda image_id: calls.append(image_id)
+        ml_tasks.embed_image, "delay", lambda image_id: calls.append(image_id)
     )
 
     img = ImageRecord(
@@ -38,3 +38,35 @@ async def test_backfill_enqueues_missing(db, monkeypatch):
     count = ml_tasks.backfill()
     assert count >= 1
     assert img.id in calls
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_backfill_respects_cpu_embed_toggle(db, monkeypatch):
+    """B3: with cpu_embed_enabled off (agent-equipped stack, no ml-worker),
+    the CPU backfill is a no-op — the GPU 'embed' backfill owns whole-image
+    embeds there. Same gate the import hooks consult before dispatching."""
+    from sqlalchemy import update
+
+    from backend.app.models import ImageRecord, MLSettings
+    from backend.app.tasks import ml as ml_tasks
+
+    calls = []
+    monkeypatch.setattr(
+        ml_tasks.embed_image, "delay", lambda image_id: calls.append(image_id)
+    )
+    db.add(ImageRecord(
+        path="/images/o.jpg", sha256="o" * 64, size_bytes=1,
+        mime="image/jpeg", width=1, height=1,
+        origin="imported_filesystem", integrity_status="unknown",
+        siglip_embedding=None,
+    ))
+    await db.execute(
+        update(MLSettings).where(MLSettings.id == 1)
+        .values(cpu_embed_enabled=False)
+    )
+    await db.commit()
+
+    assert ml_tasks.cpu_embed_enabled() is False
+    assert ml_tasks.backfill() == 0
+    assert calls == []