feat(alarms): structural degraded-status signal for truncated alarm snapshots

The truncation-cliff fix made alarm transitions truncation-safe but silent:
when GetXmlCurrentAlarms2 returns exactly maxAlmCnt records the worker
suppresses absence-implies-Clear inference and says so only in a rate-limited
stderr warning. No client and no operator could tell a complete active set
from a capped one.

Two additive proto3 booleans carry the verdict out:

- QueryActiveAlarmsReplyPayload.snapshot_truncated = 2 (worker IPC reply)
- ActiveAlarmSnapshot.from_truncated_snapshot = 16 (per record)

The per-record field is not an aesthetic choice. QueryActiveAlarms returns a
bare `stream ActiveAlarmSnapshot` with no envelope, header, or trailer, so a
per-record boolean is the only carrier that stays wire-compatible; an envelope
message would change every existing client's stream element type. The reply
payload states it too because a prefix filter can leave zero records and a
truncated fetch with nothing to report still has to say so. The flag means
"this set may be incomplete", never "this record is unreliable" — it is
independent of the subtag-fallback `degraded` field.

Detection is deliberately UNCHANGED: IsTruncatedFetch remains
`fetchedRecordCount >= maxAlarmsPerFetch`. The live probe (docs/AlarmProbeFindings.md,
ce5d8ae) could not verify whether ALARM_RECORDS/@COUNT reports the total active
count or only the records in the reply, so @COUNT is not parsed for detection;
switching to it stays blocked on probe evidence. The probe's comment
annotations in WnWrapAlarmConsumer.cs are preserved.

Reset semantics: not latched. WnWrapAlarmConsumer.FoldFetch replaces the
verdict on every poll under the same lock as the snapshot merge, so the first
sub-cap fetch clears it; GatewayAlarmMonitor.ClearCache drops it with the cache
generation it describes. A caveat that never turns off is one operators learn
to ignore.

Flow: WnWrapAlarmConsumer.LastSnapshotTruncated -> AlarmDispatcher (stamps every
record) / IAlarmCommandHandler (payload) -> MxAccessCommandExecutor reply ->
GatewayAlarmMonitor._snapshotTruncated -> IGatewayAlarmService.SnapshotTruncated
-> DashboardAlarmQueryResult -> AlarmsPage warning banner (render-side only; the
poll loop and DisposeAsync drain are untouched). The public QueryActiveAlarms
RPC forwards worker snapshots unmodified, so the per-record flag needed no
mapper change — a test pins that.

Parity: this describes OUR fetch mechanics — additive gateway metadata — not
MXAccess provider behavior. No event is synthesized and no MXAccess-observable
semantics change, so it is not a parity deviation.

Tests: worker LastSnapshotTruncated set/reset/consecutive-burst (windev-run);
gateway end-to-end truncated reply -> monitor -> public stream, with the
complete-reply control as the load-bearing assertion; AlarmsPage banner
present/absent. Docs: gateway.md alarm surface, docs/DesignDecisions.md entry.

src/ZB.MOM.WW.MxGateway.Worker/MxAccess/WnWrapAlarmConsumer.cs

@@ -93,6 +93,7 @@ public sealed class WnWrapAlarmConsumer : IMxAccessAlarmConsumer
 
     private long lastTruncationWarningMilliseconds = -TruncationWarningIntervalMilliseconds;
     private long truncatedFetchCount;
+    private bool lastSnapshotTruncated;
     private wwAlarmConsumerClass? client;
     private wwAlarmConsumerClass? ackClient;
     private bool subscribed;
@@ -125,6 +126,23 @@ public sealed class WnWrapAlarmConsumer : IMxAccessAlarmConsumer
             : DefaultMaxAlarmsPerFetch;
     }
 
+    /// <summary>
+    ///     COM-free construction, for exercising the retained-snapshot state
+    ///     machine (<see cref="FoldFetch"/> / <see cref="LastSnapshotTruncated"/>
+    ///     / <see cref="SnapshotActiveAlarms"/>) on a machine without AVEVA
+    ///     installed. <see cref="Subscribe"/> throws and <see cref="PollOnce"/>
+    ///     no-ops on an instance built this way — both need the wnwrap coclass,
+    ///     which cannot be instantiated on the macOS/Linux test matrix. Internal
+    ///     rather than public so it cannot be reached from production wiring.
+    /// </summary>
+    /// <param name="maxAlarmsPerFetch">Maximum alarms per fetch call.</param>
+    internal WnWrapAlarmConsumer(int maxAlarmsPerFetch)
+    {
+        this.maxAlarmsPerFetch = maxAlarmsPerFetch > 0
+            ? maxAlarmsPerFetch
+            : DefaultMaxAlarmsPerFetch;
+    }
+
     /// <summary>
     ///     Resolves the per-fetch cap from the launcher-provided environment
     ///     variable. A missing, unparseable, or out-of-range value falls back
@@ -373,6 +391,18 @@ public sealed class WnWrapAlarmConsumer : IMxAccessAlarmConsumer
         }
     }
 
+    /// <inheritdoc />
+    /// <remarks>
+    ///     Read without the disposed guard <see cref="SnapshotActiveAlarms"/>
+    ///     carries: this is degraded-status metadata a reply builder stamps
+    ///     alongside a snapshot, and throwing from it would fail a query whose
+    ///     snapshot half succeeded.
+    /// </remarks>
+    public bool LastSnapshotTruncated
+    {
+        get { lock (syncRoot) { return lastSnapshotTruncated; } }
+    }
+
     /// <summary>
     ///     Sink for the rate-limited truncated-fetch warning. Defaults to
     ///     <see cref="Console.Error"/>, the stream
@@ -413,14 +443,8 @@ public sealed class WnWrapAlarmConsumer : IMxAccessAlarmConsumer
         // docs/AlarmProbeFindings.md.)
         bool truncated = IsTruncatedFetch(fetchedRecordCount, maxAlarmsPerFetch);
 
-        IReadOnlyList<MxAlarmTransitionEvent> transitions;
-        int retainedCount;
-        lock (syncRoot)
-        {
-            transitions = ComputeTransitions(latestSnapshot, next);
-            ApplySnapshotUpdate(latestSnapshot, next, truncated);
-            retainedCount = latestSnapshot.Count;
-        }
+        IReadOnlyList<MxAlarmTransitionEvent> transitions =
+            FoldFetch(next, truncated, out int retainedCount);
 
         if (truncated)
         {
@@ -436,6 +460,37 @@ public sealed class WnWrapAlarmConsumer : IMxAccessAlarmConsumer
         }
     }
 
+    /// <summary>
+    ///     Folds one fetch into the retained state under a single lock: the
+    ///     transition diff, the snapshot merge/replace, and the truncation
+    ///     verdict move together. Splitting them would let a concurrent
+    ///     <see cref="SnapshotActiveAlarms"/> / <see cref="LastSnapshotTruncated"/>
+    ///     pair read a capped snapshot alongside the previous poll's "complete"
+    ///     verdict — precisely the false all-clear the signal exists to prevent.
+    ///     The verdict is replaced, never latched: a sub-cap fetch is complete
+    ///     and restores absence authority, so leaving the flag set would strand
+    ///     the operator banner on after a single burst.
+    /// </summary>
+    /// <param name="next">The snapshot just parsed from the fetch.</param>
+    /// <param name="truncated">Whether the fetch hit the per-fetch cap.</param>
+    /// <param name="retainedCount">Size of the retained snapshot after the fold.</param>
+    /// <returns>The transitions the fetch implies.</returns>
+    internal IReadOnlyList<MxAlarmTransitionEvent> FoldFetch(
+        Dictionary<Guid, MxAlarmSnapshotRecord> next,
+        bool truncated,
+        out int retainedCount)
+    {
+        lock (syncRoot)
+        {
+            IReadOnlyList<MxAlarmTransitionEvent> transitions =
+                ComputeTransitions(latestSnapshot, next);
+            ApplySnapshotUpdate(latestSnapshot, next, truncated);
+            lastSnapshotTruncated = truncated;
+            retainedCount = latestSnapshot.Count;
+            return transitions;
+        }
+    }
+
     /// <summary>
     ///     Decides whether a fetch that came back holding
     ///     <paramref name="fetchedRecordCount"/> records hit the cap.