HBASE-30238 Preserve bulkload descriptor cluster ids for replication

Keep using BulkLoadDescriptor cluster ids when invoking HFileReplicator so replicated bulkload RPCs do not treat the target cluster as already handled. The ZooKeeper event tracking identity remains unchanged.

Author: juke-mini666

hbase-server/src/main/java/org/apache/hadoop/hbase/replication/regionserver/ReplicationSink.java

@@ -317,8 +317,7 @@ public void replicateEntries(List<WALEntry> entries, final ExtendedCellScanner c
               }
               // Map of table name Vs list of pair of family and list of
               // hfile paths from its namespace
-              List<String> clusterIds = entry.getKey().getClusterIdsList().stream()
-                .map(k -> toUUID(k).toString()).collect(Collectors.toList());
+              List<String> clusterIds = bld.getClusterIdsList();
               Map<String, List<Pair<byte[], List<String>>>> bulkLoadHFileMap =
                 bulkLoadsPerClusters.computeIfAbsent(clusterIds, k -> new HashMap<>());
               buildBulkLoadHFileMap(bulkLoadHFileMap, table, bld);