Set nx and ny to 1 for single column

[billsacks]

Resolves #1622 - see that issue for details.

[billsacks]

I am not totally confident in this fix, so would be happy to have someone else look at this and think about whether this seems like the right thing to do.

@mvertens - FYI, and I'd welcome your eyes on this, too.

[billsacks]

I ran two tests with comparisons against baselines:

SCT_D_Ln7.ne3_ne3_mg37.QPC5.derecho_intel.cam-scm_prep
SCT_D_Ln7.ne3_ne3_mg37.QPC6.derecho_intel.cam-scm_prep_c6

Both tests also included the changes in ESCOMP/CDEPS#422 and ESCOMP/CMEPS#679, but I don't think those changes are needed.

Tests passed and baseline comparisons passed.

I also manually compared the CAM and CPL hist files from case2 (the single-column case) in these two tests against baselines. CAM hist files were bit-for-bit. CPL hist files had different dimension sizes, as expected / intended: the new test has dimension sizes of 1 for all variables in the cpl hist files. I created versions of the baseline cpl hist files where I extracted just the first point for each dimension and then redid the comparison; this comparison was bit-for-bit, as hoped for.

[billsacks]

@jtruesdal - FYI