Make SnapshotHistoryStore retry indexing history data on EsRejectedExecutionException

ILMHistoryStore uses BulkProcessor to retry indexing history up to 3 times if it hits a EsRejectedExecutionException because the cluster was too busy. From what I can tell, SnapshotHistoryStore just makes a single attempt, and logs an error and moves on if it hits EsRejectedExecutionException. I have seen this several times in production logs. We ought to have SnapshotHistoryStore use BulkProcessor to make a few attempts before giving up.