Add pipeline/workflow capabilities to the indexer orchestrator
Adding dependency between indexers, so that an indexer run gets results from a run of another indexer, without doing a roundtrip to the database.
Migrated from T1256 (view on Phabricator)