sedispatch: improve performance using cache friendly api
    
    It turns out that using auparse_goto_record_num() is not cache friendly.
    Since it is only processing AVC events, there is no chance of seeking
    around and missing the AVC record. So, that part of the program is
    switched out to use auparse_next_record() which only moves through the
    event once.
    
    Also unused variables were remove and the loop simplified.
    
    This change gets about 9% more speed. For reference, this
    is how I checked the speed:
    
    time ./sedispatch < /var/log/audit/audit.log >/dev/null