Job array dependency based upon state is now dependent upon the state of
the array as a whole (e.g. afterok requires ALL tasks to complete
sucessfully, afternotok is true if ANY tasks does not complete successfully,
and after requires all tasks to at least be started).

Add functions to test the state of a job array, for example
test_job_array_pending() returns true if ANY task of a job array
is pending and test_job_array_completed() returns true if ALL
tasks of a job array are completed.

spaces.

Count is reset running array tasks on reconfig

Recommend use of job arrays and multiple job steps per job.

Cray needs for task_g_post_step() to be called before resetting
the CPU frequency. We also need to reset CPU frequency before
notifying srun of task completion. Logic reorganized to satisfy
this requirement
see bug 1011

sched/backfill - Set expected start time of job submitted to multiple
partitions to the earliest start time on any of the partitions.
Previous logic would set the time to that of the last partition
tested.

This issue was addressed differently and the original 1 second
sleep can be restored for easier performance comparison with
Slurm verstion 14.03.

Previously job would only run in first listed partition.

Fix gang scheduling for jobs submitted to multiple partitions.
Previous logic assumed the job's "partition" field contained a
single partition name, that in which the job is running. That
was recently changed in order to support job's being requeued,
which we want to be runable in all of it's valid partitions.

backport of commit 9b4f3634

Previously the --partition option would not work with the new
--priority option for jobs submitted to multiple partitions.

Added squeue -P/--priority option that can be used to display pending jobs
in the same order as used by the Slurm scheduler even if jobs are submitted
to multiple partitions (job is reported once per usable partition).

Modify job array logic to properly support the MaxJobCount
configuration parameter with the new job array data structure.

This part handles the decrement portion of logic.

Rename the function local variable "job_count" in order to avoid
confusion with the global variable by the same name.
No change to logic.

NFS file system delays was causing the test for file existence
to periodically fail. Adding an "ls" call syncs the file system
and fixes the problem.