Fix daemon shutdown hang and concurrent-send race

- daemon_net: close the client transport on context cancellation so the
  per-connection Recv loop unblocks; otherwise wg.Wait() in the accept loop
  hung on a still-connected client and the daemon never stopped.
- protocol: guard ConnTransport.Send with a mutex so the subscriber pump and
  command handlers can push frames concurrently without racing the bufio.Writer.

Fixes TestDaemonDetachReattachPreservesProcess (now passes under -race).

internal/app/daemon_net.go

@@ -232,6 +232,14 @@ func handleDaemonAttach(ctx context.Context, registry *ProjectRegistry, t protoc
 		_ = sendSnapshot(t, project, view.FocusedID)
 	}
 
+	// Close the transport when the daemon context is cancelled (shutdown or
+	// `daemon stop`). Without this the t.Recv() loop below blocks forever on a
+	// still-connected client and the accept loop's wg.Wait() never returns.
+	go func() {
+		<-ctx.Done()
+		_ = t.Close()
+	}()
+
 	done := make(chan struct{})
 	go func() {
 		defer close(done)