tests: retry on timeout in partition balancer test

In partition balancer maintenance mode test one of the nodes is made
unavailable. In some cases the node may be suspended while being a
`redpanda/controller` partition leader. In this case changing
configuration may result in timeout, which is perfectly fine. We should
allow Admin client to retry in this case.

Fixes: redpanda-data#14047

Signed-off-by: Michal Maslanka <michal@redpanda.com>

tests/rptest/tests/partition_balancer_test.py

@@ -777,7 +777,7 @@ def test_maintenance_mode(self, kill_same_node):
         node_id = self.redpanda.idx(node)
 
         rpk = RpkTool(self.redpanda)
-        admin = Admin(self.redpanda)
+        admin = Admin(self.redpanda, retry_codes=[503, 504])
 
         rpk.cluster_maintenance_enable(node, wait=True)
         # the node should now report itself in maintenance mode