fix(communication): DebugStreamService drops the session on an InstanceNotFound snapshot (review 2026-10-21 03/M-3)

src/ZB.MOM.WW.ScadaBridge.Communication/DebugStreamService.cs

@@ -207,6 +207,19 @@ public class DebugStreamService : IDisposable
                 ex);
         }
 
+        if (snapshot.InstanceNotFound)
+        {
+            // The bridge actor already stopped itself on the not-found snapshot. Do not keep a
+            // live-looking session for it (review 2026-10-21 03/M-3): a consumer that never
+            // calls StopStream (the SignalR hub just forwards the flag) left the row in the
+            // registry until disconnect, and every keepalive round Telling a dead actor.
+            _sessions.TryRemove(sessionId, out _);
+            _logger.LogInformation(
+                "Debug stream for {Instance} on site {Site} not started: instance is not deployed there",
+                instanceUniqueName, siteIdentifier);
+            return new DebugStreamSession(sessionId, snapshot);
+        }
+
         _logger.LogInformation("Debug stream {SessionId} started for {Instance} on site {Site}",
             sessionId, instanceUniqueName, siteIdentifier);
 
@@ -241,6 +254,9 @@ public class DebugStreamService : IDisposable
         }
     }
 
+    /// <summary>Test/diagnostic accessor: the number of registered (attached) sessions.</summary>
+    internal int SessionCount => _sessions.Count;
+
     /// <summary>
     /// Sends one keepalive round to every attached session. Exposed (internal) so tests can
     /// drive the keepalive deterministically instead of waiting on the timer.