Fix flakyness in failure_split_cleanup (#6450)

Sometimes in CI our failure_split_cleanup test would fail like this:
```diff
     CALL pg_catalog.citus_cleanup_orphaned_resources();
-NOTICE:  cleaned up 79 orphaned resources
+NOTICE:  cleaned up 82 orphaned resources
     SELECT operation_id, object_type, object_name, node_group_id, policy_type
```

Source:
https://app.circleci.com/pipelines/github/citusdata/citus/28107/workflows/4ec712c9-98b5-4e90-9806-e02a37d71679/jobs/846107

The reason was that previous tests in the schedule would also create
some orphaned resources. Sometimes some of those would already be
cleaned up by the maintenance daemon, resulting in a different number of
cleaned up resources than expected. This cleans up any previously
created resources at the start of the test without logging how many
exactly were cleaned up. As a bonus this now also allows running this
test using check-failure-base.

src/test/regress/expected/failure_split_cleanup.out

@@ -14,6 +14,10 @@ SET citus.next_cleanup_record_id TO 11;
 SET citus.shard_count TO 2;
 SET citus.shard_replication_factor TO 1;
 SELECT pg_backend_pid() as pid \gset
+-- cleanup any leftovers from previous tests so we get consistent output
+SET client_min_messages TO WARNING;
+CALL pg_catalog.citus_cleanup_orphaned_resources();
+RESET client_min_messages;
 -- Disable defer shard delete to stop auto cleanup.
 ALTER SYSTEM SET citus.defer_shard_delete_interval TO -1;
 SELECT pg_reload_conf();
@@ -94,7 +98,7 @@ ERROR:  connection to the remote node localhost:xxxxx failed with the following
 
     \c - postgres - :master_port
     CALL pg_catalog.citus_cleanup_orphaned_resources();
-NOTICE:  cleaned up 79 orphaned resources
+NOTICE:  cleaned up 4 orphaned resources
     SELECT operation_id, object_type, object_name, node_group_id, policy_type
     FROM pg_dist_cleanup where operation_id = 777;
  operation_id | object_type | object_name | node_group_id | policy_type
@@ -683,6 +687,6 @@ CONTEXT:  while executing command on localhost:xxxxx
 
 -- Cleanup
 \c - postgres - :master_port
+SET client_min_messages TO WARNING;
 DROP SCHEMA "citus_failure_split_cleanup_schema" CASCADE;
-NOTICE:  drop cascades to table citus_failure_split_cleanup_schema.table_to_split
 -- Cleanup

src/test/regress/sql/failure_split_cleanup.sql

@@ -16,6 +16,11 @@ SET citus.shard_count TO 2;
 SET citus.shard_replication_factor TO 1;
 SELECT pg_backend_pid() as pid \gset
 
+-- cleanup any leftovers from previous tests so we get consistent output
+SET client_min_messages TO WARNING;
+CALL pg_catalog.citus_cleanup_orphaned_resources();
+RESET client_min_messages;
+
 -- Disable defer shard delete to stop auto cleanup.
 ALTER SYSTEM SET citus.defer_shard_delete_interval TO -1;
 SELECT pg_reload_conf();
@@ -304,5 +309,7 @@ SELECT create_distributed_table('table_to_split', 'id');
 
 -- Cleanup
 \c - postgres - :master_port
+
+SET client_min_messages TO WARNING;
 DROP SCHEMA "citus_failure_split_cleanup_schema" CASCADE;
 -- Cleanup