Runtime fail due to timeout in fates PRT2 test on izumi

[adrifoster]

In creating fates baselines for ctsm5.1.dev116 one of the nag compiler tests on izumi failed:

ERS_D_Ld30.f45_f45_mg37.I2000Clm50FatesCruRsGs.izumi_nag.clm-FatesColdPRT2

This text ran fine on cheyenne with the intel compiler but times out (I tried letting it go up to 4 hours...) on Izumi. It seems to be hanging at an initial history output writing.

@glemieux and I were trying to troubleshoot but in the interest of getting my CTSM-side PR in I am making an issue here so we can add it to the expected fails list. @glemieux thinks it may be one of the multi-plexed history variables that is causing the issue so we have a plan to try to figure out which one it might be.

[slevis-lmwg]

Just a note that the same test has now failed for me twice in the CREATE_NEWCASE phase as posted here:
ESCOMP/CTSM#2052 (comment)

Submitting manually with ./create_test gets back to the original failure in the RUN phase.