Normally the ID of the report is read via the underlying file. With a CSV provider this is not possible, because in a csv file you have no possibility to specify it in a structured way. Therefore the CSV provider needs an ID to tag the read report and to find reports of past builds to create the history.