the Orchestrator needs array of entry csv files for distributing entry data to multiple executors.
That is inconvenient than one single csv file contains large size of entries.
Perhaps we can handle this issue with creating a kind of entry loader trait.