test(analysis): assert Cancelled event is published; correct cron-test rationale (0.87.21)

0.87.13 followup. The Cancelled publish had no asserting test —
mechanical revert of the publish block shipped green. New sqlx test
subscribes before spawn, drives SlowDispatcher into flight, cancels,
asserts both Started and Cancelled frames with breadcrumb fields.
Mutation-confirmed.

Also corrects: cron-test rationale comment (inverted unlock-axis),
Cancelled docstring (self-contradicting clear-triggers sentence).

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

backend/tests/crawler_daemon.rs

@@ -896,11 +896,12 @@ async fn cron_shutdown_does_not_block_on_in_flight_metadata_pass(pool: PgPool) {
     );
 
     // The advisory lock must be released so another replica's cron can
-    // pick up the slot. Without this assertion, a refactor that moved
-    // the `pg_advisory_unlock` under the cancel arm (instead of running
-    // unconditionally below the select!) would ship green — only to
-    // wedge multi-replica deployments where the second replica's cron
-    // would block on `pg_try_advisory_lock` forever. `pg_try_advisory_lock`
+    // pick up the slot. The 0.87.4 fix runs `pg_advisory_unlock`
+    // UNCONDITIONALLY below the `tokio::select!` — both the
+    // cancel-arm and the body-completed paths converge there. Without
+    // this assertion, a refactor that moved the unlock INTO the
+    // body-completed arm only (skipping the cancel path) would ship
+    // green under shutdown-latency-only coverage. `pg_try_advisory_lock`
     // returns true iff the lock is free, so we use it as the probe.
     let acquired: bool = sqlx::query_scalar("SELECT pg_try_advisory_lock($1)")
         .bind(CRON_LOCK_KEY)