Avoid rolling over twice in searchable snapshot test (#108790)

Manually rolling over the data stream was 1. not necessary, and
2. causing issues when it happened after ILM had already rolled
over, which would make ILM convert the second backing index into a
searchable snapshot as well, possibly preventing the data stream
from being deleted.

x-pack/plugin/ilm/qa/multi-node/src/javaRestTest/java/org/elasticsearch/xpack/ilm/actions/SearchableSnapshotActionIT.java

@@ -851,13 +851,16 @@ public void testSearchableSnapshotsInHotPhasePinnedToHotNodes() throws Exception
             )
         );
 
-        indexDocument(client(), dataStream, true);
-        String firstGenIndex = DataStream.getDefaultBackingIndexName(dataStream, 1L);
+        // Create the data stream.
+        assertOK(client().performRequest(new Request("PUT", "_data_stream/" + dataStream)));
+
+        var backingIndices = getBackingIndices(client(), dataStream);
+        String firstGenIndex = backingIndices.get(0);
         Map<String, Object> indexSettings = getIndexSettingsAsMap(firstGenIndex);
         assertThat(indexSettings.get(DataTier.TIER_PREFERENCE), is("data_hot"));
 
         // rollover the data stream so searchable_snapshot can complete
-        rolloverMaxOneDocCondition(client(), dataStream);
+        indexDocument(client(), dataStream, true);
 
         final String restoredIndex = SearchableSnapshotAction.FULL_RESTORED_INDEX_PREFIX + firstGenIndex;
         assertBusy(() -> {