fix(ingest): zero queue-depth gauge on all-queues-drained (review round 4)

- metrics: update_ingest_queue_depth guarded on `not by_queue`, which conflated
  None (memory backend no-op) with {} (postgres, ALL queues drained). When every
  queue drains at once, get_ingest_job_counts_by_queue returns {} and the
  pre-zero loop was skipped, leaving a stale ghost backlog in the gauge. Guard on
  `by_queue is None` only; add an all-drained regression test.
- procrastinate: note that INGEST_TRANSIENT_MAX_ATTEMPTS is snapshotted at
  blueprint-build time (restart to pick up changes).

Deck #323.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

tests/unit/test_ingest_queue_depth_metric.py

@@ -39,3 +39,17 @@ def test_none_is_noop(metric_sample):
     assert metric_sample(
         _METRIC, {"queue": "ingest-fast", "status": "doing"}
     ) == approx(2)
+
+
+def test_all_queues_drained_empty_dict_zeroes(metric_sample):
+    # postgres backend with every queue drained → get_ingest_job_counts_by_queue
+    # returns {} (list_queues_async drops empty queues). An empty dict is NOT the
+    # memory-backend no-op: it must still zero every managed queue's gauge.
+    update_ingest_queue_depth({"ingest-fast": {"todo": 9}})
+    assert metric_sample(_METRIC, {"queue": "ingest-fast", "status": "todo"}) == approx(
+        9
+    )
+    update_ingest_queue_depth({})
+    assert metric_sample(_METRIC, {"queue": "ingest-fast", "status": "todo"}) == approx(
+        0
+    )