Spark\'s Broadcast join will broadcast copy of data to each executor. Once the join operation is complete shouldn\'t we unpersist the copy on each executor? Or does spark handle
