chore(followups): reviewer-recommended tests, comments, and hardening from the remediation reviews

The remediation reviews approved every task but left a tail of small notes.
This lands the gateway-side half of them.

Hardening (behavior changes, all narrow):

- BuildFilteredWriteBulkCommand's unreachable default case failed OPEN: a
  fifth bulk-write kind added upstream without a filter case here would have
  shipped the DENIED entries to the worker while reporting them denied to the
  caller. It now throws UnreachableException.
- SqliteCanonicalAuditStore.ListRecentAsync no longer throws on a row it
  cannot date. The retention sweep deliberately preserves such rows (SQLite's
  datetime() yields NULL, so the DELETE never matches), which guaranteed the
  dashboard's recent-audit view would meet one eventually and lose the whole
  page to it. The row is now reported at DateTimeOffset.MinValue with every
  other column intact, behind an optional logger.
- The audit drain loop's finally now completes the channel writer alongside
  detaching the drain, so a producer that raced past the attached check takes
  the write-through branch instead of stranding its event in a buffer nobody
  reads until shutdown. TryComplete is idempotent, so StopAsync is unaffected.

Tests:

- MapCommandReply ownership (Assert.Same on the inner reply), mirroring the
  existing MapEvent ownership test.
- Redactor key-id length boundary at exactly 64 and 65 characters, pinning
  which way it fails. Nothing validates key-id length at creation, so
  docs/Diagnostics.md's "which no issued key id does" is now stated as the
  heuristic it is.
- ApiKeyFailureLimiter.Reset with a PartitionResolution whose partition was
  evicted between the Check and the Reset: inert, and clears nobody else's
  block.
- Constraint-cache concurrency stress: the cap is enforced by the inserting
  thread, so overshoot must be transient and proportional to the in-flight
  inserters, and the cache must settle at or under the cap.
- ListRecentAsync against a raw-SQL undateable row.

Comment/doc accuracy:

- EventsHubViewerRegistry.ReleaseConnection records that it relies on
  SignalR's default sequential per-connection dispatch
  (MaximumParallelInvocationsPerClient = 1).
- A PERF(followup) note on Invoke's double session resolve and why removing it
  needs a SessionManager overload.
- SessionEventDistributor: the volatile-field comment named the pump as the
  lock-free reader, but the pump's single capture point is inside _replayLock;
  the genuinely lock-free reader is SubscriberCount. OnSubscriberOverflow's
  "cannot be observed here" now excepts the DisposeAsync abandon path. The
  churn test names its ConcurrentDictionary bucket-order assumption and that a
  violation surfaces as a read timeout, not a silent pass.
- The two "restores the sequential drain's behavior" claims (SessionManager,
  docs/Sessions.md) were wrong: the sequential drain leaked too, because
  KillWorkerAsync's entry ThrowIfCancellationRequested aborted the whole loop
  on the first session for zero kills. Reworded to "fixes a leak the
  sequential drain also had", with the sweep-bound/shutdown-unbound
  ParallelOptions asymmetry explained.
- ISessionManager.ShutdownAsync's token doc: it degrades the drain to a kill
  sweep rather than cancelling it, with the bounded overrun stated.
  SessionShutdownHostedService.StopAsync records that its cancellation-logging
  branch is now unreachable.

src/ZB.MOM.WW.MxGateway.Tests/Security/Authorization/ApiKeyFailureLimiterTests.cs

@@ -414,6 +414,46 @@ public sealed class ApiKeyFailureLimiterTests
         Assert.True(limiter.IsTracked(arriving));
     }
 
+    /// <summary>
+    /// The <c>PartitionResolution</c> handed back by <c>Check</c> is carried across an await (the
+    /// inner verification) before <c>Reset</c> consumes it, so the partition it names can be evicted
+    /// in between. Applying a stale resolution must be inert — never throw, and never clear a
+    /// partition or aggregate that belongs to somebody else. The failure direction that matters is
+    /// "clears too little", which costs the caller a probe wait; "clears somebody else's block"
+    /// would be a throttle bypass.
+    /// </summary>
+    [Fact]
+    public void Reset_WithStaleResolutionForEvictedPartition_IsInertAndClearsNoOtherBlock()
+    {
+        ManualTimeProvider clock = new(DateTimeOffset.UnixEpoch);
+        ApiKeyFailureLimiter limiter = CreateLimiter(clock, limit: 3, aggregateLimit: 3, maxPartitions: 2);
+        ApiKeyThrottlePartition victim = new("ipv4:10.0.0.1:1", "victim");
+        ApiKeyThrottlePartition other = new("ipv4:10.0.0.2:1", "other");
+        ApiKeyThrottlePartition arriving = new("ipv4:10.0.0.3:1", "arriving");
+
+        RecordFailures(limiter, victim, 3);
+        Assert.Equal(
+            ApiKeyThrottleDecision.ThrottledByPeer,
+            limiter.Check(victim, out ApiKeyFailureLimiter.PartitionResolution stale));
+        Assert.True(stale.IsResolved);
+
+        // Age the victim's window out, then push the map past its cap so eviction takes it —
+        // expired windows are the first eviction preference, so this is deterministic. The
+        // resolution captured above now names a partition that no longer exists.
+        clock.Advance(Window + TimeSpan.FromSeconds(1));
+        RecordFailures(limiter, other, 3);
+        RecordFailures(limiter, arriving, 3);
+        Assert.False(limiter.IsTracked(victim));
+
+        limiter.Reset(victim, stale);
+
+        Assert.True(limiter.IsTracked(other));
+        Assert.True(limiter.IsTracked(arriving));
+        Assert.Equal(2, limiter.TrackedAggregateCount);
+        Assert.Equal(ApiKeyThrottleDecision.ThrottledByPeer, limiter.Check(other));
+        Assert.Equal(ApiKeyThrottleDecision.ThrottledByPeer, limiter.Check(arriving));
+    }
+
     private static void RecordFailures(ApiKeyFailureLimiter limiter, ApiKeyThrottlePartition partition, int count)
     {
         for (int i = 0; i < count; i++)