Disable lag detector in testAckListenerReceivesNoAckFromHangingFollower (#94758)

This test implicitly assumes that the final `stablise()` phase will end
before the hanging follower is removed from the cluster by the lag
detector, but changes elsewhere in the system may invalidate this
assumption. This commit substantially extends the lag detection timeout
to avoid this problem.

server/src/test/java/org/elasticsearch/cluster/coordination/CoordinatorTests.java

@@ -812,7 +812,13 @@ public class CoordinatorTests extends AbstractCoordinatorTestCase {
     }
 
     public void testAckListenerReceivesNoAckFromHangingFollower() {
-        try (Cluster cluster = new Cluster(3)) {
+        try (
+            Cluster cluster = new Cluster(
+                3,
+                true,
+                Settings.builder().put(LagDetector.CLUSTER_FOLLOWER_LAG_TIMEOUT_SETTING.getKey(), "1000d").build()
+            )
+        ) {
             cluster.runRandomly();
             cluster.stabilise();
             final ClusterNode leader = cluster.getAnyLeader();
@@ -832,7 +838,13 @@ public class CoordinatorTests extends AbstractCoordinatorTestCase {
             assertTrue("expected eventual ack from " + leader, ackCollector.hasAckedSuccessfully(leader));
             assertFalse("expected no ack from " + follower0, ackCollector.hasAcked(follower0));
 
+            logger.info("--> publishing final value to resynchronize nodes");
             follower0.setClusterStateApplyResponse(ClusterStateApplyResponse.SUCCEED);
+            ackCollector = leader.submitValue(randomLong());
+            cluster.stabilise(DEFAULT_CLUSTER_STATE_UPDATE_DELAY);
+            assertTrue(ackCollector.hasAckedSuccessfully(leader));
+            assertTrue(ackCollector.hasAckedSuccessfully(follower0));
+            assertTrue(ackCollector.hasAckedSuccessfully(follower1));
         }
     }