Verify that a T compset is bit-for-bit (or nearly so) with an I compset used to generate the forcing data

When we implemented or significantly changed the T compset (glc forced by dlnd) in mct/cpl7, I did a test to verify that I could get bit-for-bit the same answers (or at most roundoff-level different) from the following:

(Case 1) An I compset (evolving ice sheet in CISM coupled to CLM); this case was set up to produce cpl aux hist files needed to force dlnd in case 2

(Case 2) A T compset forced by the cpl aux hist files from Case 1

Once the cpl aux hist capability is implemented and vetted (from @mvertens ), we should do a similar set of tests with CMEPS/CDEPS.