If ADR process crashes, report status remains in "in progress" state until manually updated in db.

An issue exists in ADR generation where if the report generation java process suffers a hard failure (like an out of memory exception) on the cloud node, there is no way for the failure status to be propagated back to the database. The end result is an endless "in progress" status for that report, even though it is not running. The solution to this is the same as implemented for crawls that experience hard failures - compare the list of system processes to known report statuses and for those not running, update the the report status.