Description
Currently scan query will iterate over the whole cache even if single partition is needed.
Query should reserve partition in owning state to make sure that results are consistent on changing topology.
Query should try primary node if primary node is owner and fallback to owner backups otherwise