Fix Qt BIP78receiver trying fallback after success

Before this commit, if receiving payjoin (BIP78) via
JoinmarketQt, after the payjoin was successfully broadcast
by the sender, and the new utxo(s) recognized in the wallet,
the Qt app nevertheless attempted to broadcast the initial
payment transaction as the delayed call was not getting canceled
by the JMBIP78ReceiverManager (this attempt of course always
failed as the coins were spent, so was not dangerous, but merely
confusing). This is now fixed.

jmclient/jmclient/payjoin.py

@@ -663,7 +663,7 @@ def fallback_nonpayjoin_broadcast(err, manager):
     log.info("Payment made without coinjoin. Transaction: ")
     log.info(btc.human_readable_transaction(original_tx))
     manager.set_broadcast(False)
-    if manager.timeout_fallback_dc.active():
+    if manager.timeout_fallback_dc and manager.timeout_fallback_dc.active():
         manager.timeout_fallback_dc.cancel()
     quit()
 
@@ -731,7 +731,7 @@ def process_payjoin_proposal_from_server(response_body, manager):
         log.info("Payjoin transaction broadcast successfully.")
         # if transaction is succesfully broadcast, remove the
         # timeout fallback to avoid confusing error messages:
-        if manager.timeout_fallback_dc.active():
+        if manager.timeout_fallback_dc and manager.timeout_fallback_dc.active():
             manager.timeout_fallback_dc.cancel()
         manager.set_broadcast(True)
     if manager.mode == "command-line" and reactor.running:
@@ -1103,7 +1103,7 @@ class JMBIP78ReceiverManager(object):
             self.uri_created_callback = self.info_callback
         else:
             self.uri_created_callback = uri_created_callback
-        # This callback used by GUI as a signal that it can
+        # This callback is used by GUI as a signal that it can
         # signal the user that the dialog is close-able:
         self.shutdown_callback = shutdown_callback
         self.receiving_address = None
@@ -1201,6 +1201,13 @@ class JMBIP78ReceiverManager(object):
     def shutdown(self):
         self.tor_connection.protocol.transport.loseConnection()
         process_shutdown(self.mode)
+        # on receiver side, if we are part of a long running
+        # process (meaning above process_shutdown is a no-op),
+        # we need to abandon the delayed call (this is the normal
+        # success case):
+        tfdc = self.pj_server.manager.timeout_fallback_dc
+        if tfdc and tfdc.active():
+            tfdc.cancel()
         self.info_callback("Hidden service shutdown complete")
         if self.shutdown_callback:
             self.shutdown_callback()