Removed unnecessary count; index on submit_time
    
    As mentionted in T613, the ResulsDB is painfully slow when showing results,
    because we call `q.count()` for each query to be able to compute total number
    of results meeting the query criteria.
    This is unnecessary, thus removing it.
    
    Differential Revision: https://phab.qadevel.cloud.fedoraproject.org/D635