test(dm): escalate to SIGKILL in wait_process_exit after 120s timeout

dm-master.test occasionally hangs on SIGHUP (stuck in etcd compaction
or DDL coordination), causing cleanup_process to fail with "didn't
exit after 120 seconds" and killing the entire test group. Instead of
failing, escalate to SIGKILL so cleanup proceeds and the next test
can start.

Affects both classic and next-gen CI — seen in ha_cases, ha_cases3,
and other HA tests where multiple masters are started/stopped.

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

Author: joechenrh

dm/tests/_utils/wait_process_exit

@@ -16,5 +16,12 @@ while [ $WAIT_COUNT -lt 120 ]; do
 	((WAIT_COUNT++))
 done
 
-echo "process $process didn't exit after 120 seconds, current processlist: $(ps aux | grep $process | grep -v 'grep' | grep -v 'wait_process_exit')"
-exit 1
+echo "process $process didn't exit after 120 seconds, escalating to SIGKILL"
+ps aux | grep $process | grep -v 'grep' | grep -v 'wait_process_exit' | awk '{print $2}' | xargs -r kill -9 2>/dev/null || true
+sleep 2
+if ps aux | grep $process | grep -v 'grep' | grep -v 'wait_process_exit' >/dev/null 2>&1; then
+	echo "process $process still alive after SIGKILL"
+	exit 1
+fi
+echo "process $process killed with SIGKILL"
+exit 0