debug as well as inform the user of the limit through the user_err_msg.

Bug 4669

Bug 4322

Remove "job_id" from kill_job_on_node() as we always have valid
  "job_ptr" function argument.

The argument is redundant and possibly confusing for pack jobs

The value is redundant and possibly misleading for pack jobs

The information is redundant, and possibly inaccurate for
  heterogeneous jobs

Value of pack_offset was previously always 0, resulting in problems for
  slurmd/slurmstepd logic currently under development

The original logic was preventing a second step from starting due
  to propagating the job's memory limit, which caused the test to
  fail.

Bug 4634

job container where if the step was canceled would also cancel the stepd
erroneously.

Bug 4634

Declared variable-length array (VLA) has zero size

CLANG assumed new_cnt was <= 0.

uint64_t grp_used_tres[new_cnt],
	 grp_used_tres_run_secs[new_cnt];

No change to logic

one already exists owned by a different user will be logged and the job
held.

Bug 4614

Bug 4681

Rather than merging all of the configuration parameters from
the same plugin into a single alphabetized list, group them
by plugin type and log teh plugin name.
bug 4036

In preparation for release of v18.08.0-pre1 later this week.

The --batch option must be a subset of those in job --constraint

Bug 4683

Coverity CID 182420

For a heterogeneous job allocation with reserved MPI ports, give
  all components access to the full range of reserved ports.

Some of these variables were re-used in later code, so moving them
outside of the if logic broke things.

No change in logic

from when last started.

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

invalid reads.

Bug 4664