fix(events,ci): Phase E review fixes — DLQ structlog + CI slow-test filter

Two findings from the Phase E batched review:

1. CRITICAL: .github/workflows/pr.yml ran `pytest apps/api/tests` with no
   marker filter. Task 26's slow integration test (310s sleep for reaper
   idle-threshold validation) would block every PR CI for 5+ minutes.
   Added `-m "not slow"` to the api pytest step with an inline comment.

2. Important: dlq.py wrote to the DLQ stream silently. Added structlog
   `get_logger(__name__)` and a `logger.warning("dlq.routed", ...)` call
   inside `route()` so DLQ events surface in observability.

Tests still pass (3 passed, 1 slow deselected).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: Muzaffar-codes07

.github/workflows/pr.yml

@@ -127,7 +127,9 @@ jobs:
       - run: uv sync --all-packages --extra dev
       - run: uv run --package lockin-api python -m ruff check apps/api
       - run: uv run --package lockin-api python -m mypy apps/api/app
-      - run: uv run --package lockin-api python -m pytest apps/api/tests
+      # `slow` marker excluded: long-running integration tests (e.g., 5min sleep
+      # for reaper-idle validation) run only on schedule or manual trigger, not on PRs.
+      - run: uv run --package lockin-api python -m pytest apps/api/tests -m "not slow"
 
   python-mcp:
     needs: detect

apps/api/app/events/dlq.py

@@ -15,8 +15,11 @@
 
 from redis.asyncio import Redis
 
+from app.core.logging import get_logger
 from app.events.streams import dlq_stream
 
+logger = get_logger(__name__)
+
 MAX_DELIVERIES = 3
 
 
@@ -49,4 +52,14 @@ async def route(
             "failure_count": str(failure_count),
             "original_payload": json.dumps(original_payload),
         }
-        return str(await self._redis.xadd(dlq_stream(source_stream, consumer_group), entry))
+        msg_id = str(await self._redis.xadd(dlq_stream(source_stream, consumer_group), entry))
+        logger.warning(
+            "dlq.routed",
+            source_stream=source_stream,
+            consumer_group=consumer_group,
+            original_id=original_id,
+            error_class=type(exc).__name__,
+            failure_count=failure_count,
+            dlq_msg_id=msg_id,
+        )
+        return msg_id