SMS_Lm3_D_Mmpi-serial.1x1_brazil.I2000Clm50FatesCruRsGs.izumi_intel.clm-FatesColdHydro fails when t_soisno is initialized to 272K instead of 274K

[olyson]

A change is planned for the host model CTSM such that soil temperature for a cold start will be initialized to 272K instead of the current 274K. Purpose is to avoid locking in large amounts of deep soil carbon at high latitudes. Testing indicated a failure for SMS_Lm3_D_Mmpi-serial.1x1_brazil.I2000Clm50FatesCruRsGs.izumi_intel.clm-FatesColdHydro, caused by a divide by zero in FatesHydroWTFMod.F90 at this line:

psi = this%psi_sat*(th/this%th_sat)**(-this%beta)

where th is zero.
Please see ESCOMP/CTSM#2373 for more details.
Plan is that @rgknox has graciously agreed to provide a fix for this that @olyson will test in his current configuration. For the purposes of integrating the soil temperature change in CTSM, the test in question will be labeled as an ExpectedFail.

[olyson]

I have tested a fix provided by @rgknox and it resulted in the test passing.