Another weird edge case with date stamps in rpointer files

[mnlevy1981]

@kristenkrumhardt created a new CESM case, set RUN_TYPE=branch and also set up RUN_REFCASE and RUN_REFDATE. After she built the model and the submitted job started running, the coupler aborted immediately with

  read rpointer file = rpointer.cpl
 (esm_time_mod.F90:esm_time_clockInit) ERROR rpointer file rpointer.cpl not foun
 d

We traced the issue to DRV_RESTART_POINTER being rpointer.cpl rather than rpointer.cpl.YYYY-MM-DD-SSSSS, but should there be a mechanism to tell CIME to include REF_DATE when constructing DRV_RESTART_POINTER? Maybe the default should be to include it when CONTINUE_RUN=FALSE, but there can be a mechanism to say "no, just use rpointer.cpl instead?

We think this maybe leads to unexpected inconsistencies in hybrid runs, because similar steps but using RUN_TYPE=hybrid does not abort. Perhaps the coupler just initializes itself as a startup if DRV_RESTART_POINTER is not found in a hybrid run? Some of the components initialized from restart files, so it's a little unclear what actually happened in the hybrid case...

[billsacks]

We think this maybe leads to unexpected inconsistencies in hybrid runs, because similar steps but using RUN_TYPE=hybrid does not abort. Perhaps the coupler just initializes itself as a startup if DRV_RESTART_POINTER is not found in a hybrid run? Some of the components initialized from restart files, so it's a little unclear what actually happened in the hybrid case...

My understanding is that hybrid cases don't read a cpl restart file. At least, I think that was true with cpl7. (I think that's one of the significant differences between hybrid and branch cases.)