Fix the migration, so it deals with duplicate job uuids
    
    This is required to solve our state of database, where depchecks have
    multiple jobs with the same uuid, and in the resultsdb 2.0 the uuid is
    an unique identifier of the group.
    This is done by merging the results of each "duplicate" job to the first
    one with that specified uuid, and deleting the other (now result-less)
    jobs.
    
    Differential Revision: https://phab.qadevel.cloud.fedoraproject.org/D1054