fix(reconnect): ignore server connectionTimeout when no tracks are published

After unpublishAll(), the publisher transport idles and the server
sends LeaveRequest(resume, connectionTimeout). Previously this
triggered a needless quick reconnect cycle via the websocket path.

Skip the signal cleanup when hasPublished is false and the reason is
connectionTimeout, keeping the signal connection alive so re-publishing
can proceed without disruption.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: pblazej

Sources/LiveKit/Core/Room+SignalClientDelegate.swift

@@ -58,6 +58,13 @@ extension Room: SignalClientDelegate {
             _state.mutate { $0.nextReconnectMode = .full }
             fallthrough
         case .resume:
+            // When no tracks are published, the publisher transport may idle and
+            // the server sends connectionTimeout. This is expected after unpublishAll()
+            // and doesn't require a reconnect — the subscriber transport is still alive.
+            if !_state.hasPublished, reason == .connectionTimeout {
+                log("Ignoring resume for connectionTimeout with no published tracks")
+                return
+            }
             // Abort current attempt
             await signalClient.cleanUp(withError: error)
         case .disconnect: