[TEST] Wait for no pending operations on the index shard (#118244) (#118258)

This fixes testRetryPointInTime which on teardown is looking to assert
that the operations in the translog and in the lucene index are the
same.

Previously we didn't wait for the translog operations to be applied.
This changes `assertConsistentHistoryInLuceneIndex` to wait for the
pending operations in the translog to be applied.

Fixes #117116

(cherry picked from commit b4e852a)
Signed-off-by: Andrei Dan <[email protected]>

Author: andreidan

muted-tests.yml

@@ -320,9 +320,6 @@ tests:
 - class: org.elasticsearch.xpack.application.CohereServiceUpgradeIT
   method: testCohereEmbeddings {upgradedNodes=2}
   issue: https://github.com/elastic/elasticsearch/issues/116975
-- class: org.elasticsearch.xpack.searchablesnapshots.RetrySearchIntegTests
-  method: testRetryPointInTime
-  issue: https://github.com/elastic/elasticsearch/issues/117116
 - class: org.elasticsearch.xpack.inference.InferenceRestIT
   method: test {p0=inference/40_semantic_text_query/Query a field that uses the default ELSER 2 endpoint}
   issue: https://github.com/elastic/elasticsearch/issues/117027

test/framework/src/main/java/org/elasticsearch/snapshots/AbstractSnapshotIntegTestCase.java

@@ -127,6 +127,7 @@ protected Collection<Class<? extends Plugin>> nodePlugins() {
 
     @After
     public void assertConsistentHistoryInLuceneIndex() throws Exception {
+        internalCluster().beforeIndexDeletion();
         internalCluster().assertConsistentHistoryBetweenTranslogAndLuceneIndex();
     }