Don't close the multiprocessing queue too soon

The transaction_stop is actually for RPMCALLBACK_TRANS_STOP from rpm,
which is only the end of preparation phase. The after_complete is called
by dnf after the whole transaction completes.

Also, the queue cannot be closed at this point, because transaction
errors are written there after the transaction completes and then
TransactionProgress.quit is called, which closes the queue as well.

pyanaconda/modules/payloads/payload/dnf/transaction_progress.py

@@ -95,10 +95,9 @@ def script_start(self, item, nevra, type):
  #libdnf5.rpm.to_full_nevra_string(nevra),
  self._queue.put(('configure', nevra.get_name()))
 
- def transaction_stop(self, total):
+ def after_complete(self, total):
  log.debug("Done - %s", total)
  self._queue.put(('done', None))
- self._queue.close()
 
  def cpio_error(self, item):
  log.debug("Error - %s", item.get_package().to_string())