Signed-off-by: Danny Auble <da@schedmd.com>

if on "scontrol reconfig" when AllowNodes is manually changed using
scontrol since last slurmctld restart.

Conflicts:
	doc/man/man5/slurm.conf.5

correctly with poe when attempting to run > 32 nodes.

due to either down nodes or explicit resizing.
Generated slurmctld errors of this type:
[2013-06-04T12:43:46+06:00] error: gres/gpu: step_test 68662.4294967294 gres_bit_alloc is NULL
This is a movement of the logic introduced in commit
https://github.com/SchedMD/slurm/commit/6fff97bb77d2d88aa808c47fd7880246a0c1d090
to eliminate a memory leak.

bug 326

bug 327

These changes partially address bug 326

Our tester complained that srun --usage displays --profile=all
rather than --profile=all|none|energy|lustre|network|task.
None of the other options on usage have their possible options
displayed so I made  usage display --profile=... to be similar
Then I enhanced --help to list the possible values.
I also made the change to salloc and sbatch.
bug 325

overwritten by the parallel versions thus making it so we need handle
both cases.

we've been unable to submit jobs that request one task-per-core
when we have SMT enabled (but it would work successfully for
jobs that wanted two tasks-per-core).
I ended up tracking it down to a regression caused by commit
8127554a.
See bug 322

Thanks for the feedback! I was actually planning contributing a more complete set of documentation for the SLURM web pages but have been super busy.
Feel free to copy the current documentation. I'll try to refine it a bit next week.
It might be better to leave the link out as we might move the documentation to the "proper" CSC website at some point.

Since we don't currently track energy usage per task (only per step).
Otherwise we get double the energy.

Conflicts:
	NEWS