Increase the sleep between job end and checking accounting record
from 3 to 5 seconds so the test will work reliably.

test7.20 was always leaving vestigial batch output files and lacked
job time limits (which could leave vestigial jobs on test failure).

bug 6973

Reported as conflicting thread load operations by valgrind --tool=drd.

Bugs 6189 and 4159.

This reverts commit f3d678d4.

Reported as conflicting thread load operations by valgrind --tool=drd.

Bugs 6189 and 4159.

Bug 6783 comment 35.

Bug 6959.

Bug 6944.

Bug 6944.

This is the same because xstrdup returns null on null.
Bug 6812

No real code change.

It appears this is really what this was suppose to be anyway.

Bug 5950

Update slurm.spec and slurm.spec-legacy as well.

Bug 5680

and other_cons_res.

continuation of previous commit.

Bug 5680

In preparation for next commit.

Bug 5680

Blessed by Tim.

Still respect "cray" as the keyword here for now, but add "datawarp" as
the documented form instead.

Bug 6440.

Since the slurmd will change an older protocol version to the current
version we need to make sure if an old srun sends a message we handle
things correctly.

Otherwise non-FairTree systems will have a decay period delay before
the updated value makes it into the sum.

Bug 5202.

Usagefactor matches the documentation and now multiplies TRES time
limits and usage.

Bug 5435

This is very coarse-grained locking, but as the initial implementation
did not anticipate concurrent access this is the safest approach for now.

Bug 5638.

Rather than repeat the #ifdef block through five locations, use a common
function to handle this and print a consistent error message if there
is an issue.

From getrlimit(2):
	setrlimit() now returns with errno set to EINVAL in places that
	historically succeeded. It no longer accepts "rlim_cur =
	RLIM_INFINITY" for RLIM_NOFILE. Use "rlim_cur = min(OPEN_MAX,
	rlim_max)".

(This was causing spurious error messages to appear for sbatch/srun.)