Set the environment variable SLURM_PARTITION to the partition in
which a job is running. Set for salloc, sbatch and srun.

being scheduled to run again.  This would erase the previous job info
that was still needed to clean up the nodes from the previous job run.

Move calls to core_spec_g_set/clear from slurmd to slurmstepd.
Pass count of specialized cores from slurmd to slurmstepd.

change the core_spec_g_set and clear functions to take container
ID as an argument

Move core_spec_g_suspend() and core_spec_g_resume() from slurmd
to slurmstepd daemon per Cray request.

The function arguments are unchanged, this splits the set into
four functions.

a step using --cpu-freq completes.

Conflicts:
	NEWS
	doc/html/team.shtml

Set GPU_DEVICE_ORDINAL environment variable.

Splitting the job's shared field into whole_node and share_res
in order to support gang scheduling of jobs with exclusive node
use broke the logic used to support advanced reservations containing
specific cores.

Conflicts:
	META
	NEWS
	contribs/perlapi/libslurm/perl/Slurm.xs

Previous logic would try to pick a specific node count and on a
heterogeneous system, this would cause a problem. This change
largely reverts commit a270417b

beside the numerical values.