Was sending the response_rec for every job on a normal cluster when it
only needs to send when the allocation was from a different cluster.

The response_cluster_rec was pointing to the conf's cluster_name which
gets free'd and reallocated on a reconfig. So the cluster_name was
pointing to garbage.

so that it will only run the tests if you have check.

If size is 0, it will allocate 0 + 2. 2 for the magic and the size. But
xmalloc will then retun the pointer pointing past the newly allocated
memory (p[2]) which can cause a segfault.

burst_buffer/cray - Support default pool which is not the first pool
    reported by DataWarp and log in Slurm when pools that are added or removed
    from DataWarp.
bug 3453

Refactor the loading in of plugins.

Instead of creating static functions everytime.

This reverts commit fd690a9c.

was there for jobs and steps, but for no real reason.  I couldn't see a
reason anyway.

Add missing option for PURGE_JOB which is different than the others
and was missed on the changes made when adding the usage table
flushes.

sstat doesn't work on Cray ALPS but works on native Cray setups.

Bug 3444

There are (or at least used to be) places where an exit code of 0 was used
to indicate that the job ran successfully to completion. This patch fixes
the places where the exit_code was MAX'd out to 1 to indicate that job did
not ran successfully to completion. This MAX logic was introduced in commit
f305f099.

Bug 3444

Conflicts:
	src/slurmctld/job_mgr.c

Jobs preempted with PreemptMode=REQUEUE were incorrectly recorded as
REQUEUED in the accounting.

Bug 3444

bug 3448

i.e.

AcctGatherInterconnectType = ofed,none

Bug 3412

compatibility.

Actual code change to make it happen.  Just renaming things, nothing
really changed.