request resources that reach a 'Max' limit.

this is happening because the energy status MSR registers for Package: "MSR_PKG_ENERGY_STATUS" and DRAM: "MSR_DRAM_ENERGY_STATUS"
are counted on 32 bit ( the MSRs are 64bit wide) and they wrap around every few minutes.

for info check out the attached screenshot of the Intel's doc
or here for Intel's complete doc: http://download.intel.com/products/processor/manual/325384.pdf

user mark the state canceled instead of completed.

so it gets sent again.  This isn't a major problem since the start will
happen when the job ends, but this does make things cleaner.

you query against that with -N and -E you will get all jobs during that
time instead of only the ones running on -N.

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

to a list, which can be arbitrary size

was already in error and isn't deallocating and underlying hardware goes
bad one could get overlapping blocks in error making the code assert when
a new job request comes in.

overcommit.

overcommit.

Previously only requeued the job once