Previously if a job was dependent upon a job array and the first
element of that job array was no longer in slurmctld (completed
and the job record purged) and the slurmctld restarted or was
reconfigured, the dependent job would start rather than waiting
for all elements of the job array to complete.
bug 714

Shorten the test31.1 description for better readability of errors.

Add squeue -L/--licenses option to filter jobs by license names
bug 720

successive release requests.

list_pop_top() and list_pop_bottom() functions removed.
Quicksort produces better results in almost all cases.

Use quicksort for all priority based job sorting, which improves performance
significantly with large job counts.

squeue %F format will now print the job ID for non-array jobs.
bug 716

A variable is now unused except on Bluegene systems

Partial reversion of commit 9b1dadea

Change some sched/backfill messages to be more clear
Change some sched/builtin messages and variable names to
no longer reference "backfill".

Prevent a possible buffer overflow if the task prolog output
exceeds 4096 bytes.

Expand StdIn/Out/Err string length output by "scontrol show job" from 128
to 1024 bytes.
bug 712

See bug 701

salloc/srun, instead wait in the slurmd when a step hits a node and the
prolog is still running.

Wait for a job to begin execution prior to testing its allocated
resources to confirm proper core specialization support.

Failed to find individual jobs to resume.

priority/basic - Insure job priorities continue to decrease when jobs are
submitted with the --nice option.
bug 707

If one task of a job array can not be scheduled in the main
scheduling loop, skip the subsequent tasks of that same
job array.

This reverts commit 6cd208e0.

has ran yet or not.

Cray/ALPS system, otherwise use the slurmd to wait on the prolog to gate
the start of the step.

Fix some sinfo and squeue sorting anomalies due to differences in data
types.
See bug 706