Add ability to check persisted cache on each executor #2
Description
Currently cached directory must be either shared on local file system or be directory on HDFS. Proposal to introduce small Spark job to collect what cached data is available on each executor and distribute pull requests info accordingly without requirement on shared folder.