fix(siteeventlog): suppress snapshot-resync alarm re-emit + coverage + hardening (review)

src/ZB.MOM.WW.ScadaBridge.SiteRuntime/Actors/DeploymentManagerActor.cs

@@ -643,9 +643,18 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
     /// <see cref="ISiteEventLogger"/> on a deploy/enable/disable/delete outcome.
     /// Resolved optionally and never awaited so a logging failure cannot affect the
     /// deployment pipeline (matching the established ScriptActor/ScriptExecutionActor
-    /// pattern). Only reads the immutable <c>_serviceProvider</c> field, so it is
-    /// safe to call from the PipeTo continuations that report disable/delete
-    /// outcomes off the actor thread.
+    /// pattern).
+    /// <para>
+    /// <b>Thread-safety:</b> the disable (<see cref="HandleDisable"/>) and delete
+    /// (<see cref="HandleDelete"/>) paths call this from a
+    /// <see cref="System.Threading.Tasks.Task.ContinueWith(System.Action{System.Threading.Tasks.Task})"/>
+    /// continuation that runs on a thread-pool thread, NOT on the actor thread —
+    /// so it must touch only immutable, thread-safe state. It does: the only
+    /// field it reads is the <c>readonly _serviceProvider</c> captured at
+    /// construction (the resolved <see cref="ISiteEventLogger"/> is a process
+    /// singleton). No actor-private mutable state is referenced, which is what
+    /// makes calling it off the actor thread safe.
+    /// </para>
     /// </summary>
     private void LogDeploymentEvent(string severity, string instanceName, string message, string? details = null)
     {