Ticket 47533 logconv: some stats do not work across server restarts
    
    https://fedorahosted.org/389/ticket/47533
    Reviewed by: mreynolds (Thanks!)
    Branch: master
    Fix Description: Most of the stat hashes use connid,opid as the hash key.
    The connid,opid are not unique because the server recycles connection ids
    when the server restarts.  The solution is to use a triplet of
    $serverRestartCount,$connid,$opid
    as the hash key.  This uniquely identifies a particular log event over
    restarts.  There is also some additional cleanup that needs to be done when
    the server is reset - returning fds, marking connections as closed, etc.
    Also fixed some gmtime handling - we were doing too many conversions.
    This fix adds another command line option - -U - for a summary of search
    filters which are unindexed or have unindexed components.  The full detailed
    report is just too much data.
    Platforms tested: RHEL6 x86_64
    Flag Day: no
    Doc impact: no
    (cherry picked from commit 92ac8f48c1e5a8a049f4cfd1912fd570c628408f)
    (cherry picked from commit e38685be139c2420cc969499324405a30203122a)