Should some compsets/tests use DGLC%NOEVOLVE rather than SGLC?

We talked about this in today's CSEG meeting, but couldn't have a clear answer there. So I'm adding it here so it isn't lost. This probably needs input from both the land-ice folks as well as CAM folks.

CAM compsets use stub glacier SGLC, but we wondered if they should use DGLC%NOEVOLVE. The fully coupled compsets now use DGLC, but will be converted to using CISM.

For consistency with the fully coupled compsets, it's possible that DGLC should be used rather than SGLC. It would be a minor change to answers, and would have a small performance impact.