Add scontrol requeue option of "Incomplete" which will requeue jobs only if
    they failed to complete with an exit code of zero.
bug 6397

This is useful for job requeue.
bug 5693

Estimate the total count of GRES of a job at submit time.
This is used to reject jobs that will definitely exceed
a GRES limit rather than queuing them and rejecting them
when we finally determine their GRES count.
bug 6410

A block of code was repeated and does nothing but recompute a value

Option rejects user requests to shrink running jobs.

(Jobs using --no-kill may still shrink due to node failures.)

Bug 5716.

Only call to it was removed in the prior commit, and all the
interfaces returned "true" (even if they probably shouldn't have,
such as select/cray).

Add new SchedulerParameter of permit_job_expansion to permit expansion of
running jobs.

Replace the two calls to select_g_job_expand_allow() as all plugins return
true anyways at this point. Return ESLURM_NOT_SUPPORTED instead of
ESLURM_DEPENDENCY in one location as well to provide a better error
message back to the user.

Bug 5716.

thrown when compiling with CFLAGS="--converage"

Logs are produced by regression.py, and should not be checked in.

Fix regression in 2caf9e8c

Changes saccct, squeue, sinfo, sprio so that -l and -o
are mutually exclusive.

Bug 6438

Bug 6442

Bug 6441.

Bug 6196

is try to print a log message.  Since we are already in that lock it
will deadlock.

We document that DerivedExitCode is the highest exit code of the srun
steps, so we need to exclude the batch step. The extern step was already
excluded.

Bug 6196.

Bug 6196

Bug 6196

Bug 6196.

This means we can find batch steps.  NO_VAL isn't used in the code
to return the first one.

Bug 6196

delete_step_record.

so that squeue -s and sstat will show the batch step, and sacct will
show the batch step before the job finishes.

Bug 6196.

It turns out REQUEST_STEP_COMPLETE is never sent by the batch step.

Bug 6196

Run autogen.sh, which picked up one older uncommited change to
a comment line in contribs/perlapi/libslurmdb/Makefile.in.

The correct printf format for size_t is %zu not %lu.

This reverts commit 2707a9c0.

A lot of places rely on a zero size (or, after xcalloc conversion, count)
resulting in a NULL return, so stop xassert()'ing those cases and avoid
jumping to the unpack_error label.

Bug 6430