Attempt to fix rare race in test for MDEV-8031

The error-injection inject_mdev8031 simulates a deadlock kill in a specific
place, by setting killed_for_retry to RETRY_KILL_KILLED directly. If a real
deadlock kill triggers at the same time, it is possible for the thread to
complete its transaction retry and set rgi_slave to NULL before the real
readlock kill can complete in the background. This will cause a segfault
due to null-pointer access.

Fix by changing the error injection to do a real background deadlock kill,
which ensures that the thread will wait for any pending background kills to
complete.
Signed-off-by: Kristian Nielsen <knielsen@knielsen-hq.org>

sql/rpl_parallel.cc

@@ -856,8 +856,7 @@ retry_event_group(rpl_group_info *rgi, rpl_parallel_thread *rpt,
     thd->wait_for_commit_ptr->unregister_wait_for_prior_commit();
   DBUG_EXECUTE_IF("inject_mdev8031", {
       /* Simulate that we get deadlock killed at this exact point. */
-      rgi->killed_for_retry= rpl_group_info::RETRY_KILL_KILLED;
-      thd->set_killed(KILL_CONNECTION);
+      slave_background_kill_request(thd);
   });
 #ifdef ENABLED_DEBUG_SYNC
   DBUG_EXECUTE_IF("rpl_parallel_simulate_wait_at_retry", {