Fix IndexShardIT#testMaybeFlush (#52247)

Since #51905, we use the local checkpoint of the safe commit to 
calculate the number of uncommitted operations of a translog stats. If a
periodic flush triggered by afterWriteOperation completes before we sync
translog, then the last commit is not safe. We also need to sync
translog from Engine instead of the translog so that we can advance the
safe commit.

Relates #51905
Closes #52223

Author: dnhatn

server/src/test/java/org/elasticsearch/index/shard/IndexShardIT.java

@@ -349,8 +349,8 @@ public void testMaybeFlush() throws Exception {
             assertFalse(shard.shouldPeriodicallyFlush());
             assertThat(shard.flushStats().getPeriodic(), greaterThan(0L));
         });
+        shard.sync();
         assertEquals(0, translog.stats().getUncommittedOperations());
-        translog.sync();
         long size = Math.max(translog.stats().getUncommittedSizeInBytes(), Translog.DEFAULT_HEADER_SIZE_IN_BYTES + 1);
         logger.info("--> current translog size: [{}] num_ops [{}] generation [{}]",
             translog.stats().getUncommittedSizeInBytes(), translog.stats().getUncommittedOperations(), translog.getGeneration());
@@ -369,6 +369,7 @@ public void testMaybeFlush() throws Exception {
                 commitStats.getUserData(), flushStats.getPeriodic(), flushStats.getTotal());
             assertFalse(shard.shouldPeriodicallyFlush());
         });
+        shard.sync();
         assertEquals(0, translog.stats().getUncommittedOperations());
     }