Fix Race in testGetSnapshotsRequest (#61694)

The fact that the data node is already blocked on writing
data files did not guarantee that the cluster state that made
the data node start snapshotting is already applied on master.
This could lead to races where the get snapshots action still
runs based on a state without the snapshot in it, tripping the assertion.
Much safer to handle this by waiting on the non-blocking snapshot create
to return, which guarantees that the CS has been applied on master.

Closes #61541

Author: original-brownbear

server/src/internalClusterTest/java/org/elasticsearch/snapshots/SharedClusterSnapshotRestoreIT.java

@@ -2803,11 +2803,10 @@ public void testGetSnapshotsRequest() throws Exception {
         // take initial snapshot with a block, making sure we only get 1 in-progress snapshot returned
         // block a node so the create snapshot operation can remain in progress
         final String initialBlockedNode = blockNodeWithIndex(repositoryName, indexName);
-        ActionFuture<CreateSnapshotResponse> responseListener =
-            client.admin().cluster().prepareCreateSnapshot(repositoryName, "snap-on-empty-repo")
+        client.admin().cluster().prepareCreateSnapshot(repositoryName, "snap-on-empty-repo")
                 .setWaitForCompletion(false)
                 .setIndices(indexName)
-                .execute();
+                .get();
         waitForBlock(initialBlockedNode, repositoryName, TimeValue.timeValueSeconds(60)); // wait for block to kick in
         getSnapshotsResponse = client.admin().cluster()
                                    .prepareGetSnapshots("test-repo")
@@ -2816,7 +2815,6 @@ public void testGetSnapshotsRequest() throws Exception {
         assertEquals(1, getSnapshotsResponse.getSnapshots("test-repo").size());
         assertEquals("snap-on-empty-repo", getSnapshotsResponse.getSnapshots("test-repo").get(0).snapshotId().getName());
         unblockNode(repositoryName, initialBlockedNode); // unblock node
-        responseListener.actionGet(TimeValue.timeValueMillis(10000L)); // timeout after 10 seconds
         client.admin().cluster().prepareDeleteSnapshot(repositoryName, "snap-on-empty-repo").get();
 
         final int numSnapshots = randomIntBetween(1, 3) + 1;