With the new series store, we should be able to intersection series IDs as we receive them index, and then issues the fetch for chunks earlier. Hopefully this will cause the various stages (index lookup, chunk fetch) to overlap and reduce query time.