Make sure there are no duplicates on the nodes list

This is probably more of a CI bug than anything else - in CI we use
a lower tick value, which increases the odds of the periodic
repair triggering very early.

Failure:
https://github.com/rabbitmq/rabbitmq-server/actions/runs/13013004042/job/36295172214?pr=13050
(cherry picked from commit 198d8f93d3)

deps/rabbit/src/rabbit_quorum_queue.erl

@@ -1353,7 +1353,7 @@ add_member(Q, Node, Membership, Timeout) when ?amqqueue_is_quorum(Q) ->
                     Fun = fun(Q1) ->
                                   Q2 = update_type_state(
                                          Q1, fun(#{nodes := Nodes} = Ts) ->
-                                                     Ts#{nodes => [Node | Nodes]}
+                                                     Ts#{nodes => lists:usort([Node | Nodes])}
                                              end),
                                   amqqueue:set_pid(Q2, Leader)
                           end,