Add job record qos field and partition record allow_qos field.

This patch moves the QOS update of an existing job to be before the
partition update.  This ensures a new QOS value is the value used when
doing validations against things like a partition's AllowQOS and DenyQOS.

Currently if a two partitions have AllowQOS that do not share any QOS,
the order of updates prevents a job from being moved from one partition
to another using something like the following:

scontrol update job=<jobID> partition=<new part> qos=<new qos>

value for booleans.

Conflicts:
	src/common/slurm_step_layout.c

Correct task layout with CR_Pack_Node option and more than 1 CPU per task.
Previous logic would place one task per CPU launch too few tasks.
bug 1781

Backfill scheduler now considers OverTimeLimit and KillWait configuration
parameters to estimate when running jobs will exit. Initially the job's
end time is estimated based upon it's time limit. After the time limit
is reached, the end time estimate is based upon the OverTimeLimit and
KillWait configuration parameters.
bug 1774

Backfill scheduler: The configured backfill_interval value (default 30
    seconds) is now interpretted as a maximum run time for the backfill
    scheduler. Once reached, the scheduler will build a new job queue and
    start over, even if not all jobs have been tested.
bub 1774

of the size returned.

This check is redundant though since getgrouplist will return -1 if it
tries to use more than ngroups_max.  Perhaps we should just take the check
out.

No change in logic

Original patch from LLNL assumed Munge was installed, which would
  result in a build error if the Munge development package was
  not installed

If a user runs sreport and specifies a time period that stretches into
the future, this patch sets the end of the period to the current time.

In order to successfully remove the freezer cgroup at the end of
a job step, the slurmstepd process itself must first be moved
outside of the cgroup, or removal will always fail.

This fix moves the slurmstepd back to the root cgroup just
before the rmdir operations are attempted.

If the job prolog is running we can't send a signal to job step
tasks, so return SLURM_FAILURE instead of ESLURM_INVALID_JOB_ID.
This should cause the caller to retry, instead of assuming the
job step is not running on the node.

If a job step request comes in while the slurm prolog is running,
slurmd will happily launch the job step. This means that a user
could run code before the prolog is complete, which could cause
strange errors or in some cases a security issue. Instead return
an error (EINPROGRESS for now) and do not allow job steps to run
during prolog.

Create a new function, _prolog_is_running(), to determine if a job
prolog is currently running from within slurmd, and use this new
function in place of list_find_first in _wait_for_job_running_prolog.

Some cgroup code would retry continuously if somehow a cgroup
didn't exist when xcgroup_delete() was called (See for instance
proctrack/cgroup's slurm_container_plugin_wait()). To fix this
for all callers, return SUCCESS from xcgroup_delete() if rmdir(2)
return ENOENT, since the cgroup is already deleted.

Test of gres.conf configurations options
test 913