sched/backfill - Delay backfill scheduler for completing jobs only if
    CompleteWait configuration parameter is set (make code match documentation).

moved code to common location.

Addition to b9668b2b

Bug 2030

Invalid user input could generate slurmctld error(), changed to info()

Eliminated redundant data structure
Minor performance improvements

correctly when PriorityFlags=CALCULATE_RUNNING is set.

Previously the slurmctld could seg fault if the tres_alloc_str is NULL.

Prevent slurmdbd divide by zero if no associations defined at rollup time.

else behind.

uint64_t.

removes the kill_extern_procs() function added in the referenced commit.

This appears to be the way we do it with other steps and seems to work
correctly.

as all the associations from the database will be lower case.

sched/backfill: This change moves some logic used to support the
   bf_min_age_reserve configuration parameter for slightly better
   performance.

Correct job task count calcuation if only node count and ntasks-per-node
    options supplied.
bug 2196

bug 2199

SchedulerParameters now accepts keyword of "max_rpc_count" in
  addition to the documented "max_rpc_cnt"

sched/backfill: If max_rpc_cnt is configured and the backlog of RPCs has
    not cleared after yielding locks, then continue to sleep.

requesting any specific association.

calls.

snowflake10 after snowflake8

Disable srun --bcast option test if SlurmdUser != root as the
  operation/test  will not work otherwise.

This would have helped to identify the performance issue associated
with bug 2174

sacctmgr output used a suffix of "PU" for per user, but only
  accepted input with a suffix of "PerUser". The output all
  used a suffix of "PU". The code has been modified to accept
  both sufficies (e.g. MaxJobsPerUser=# or MaxJobsPU=#).