Glad's is_in_active_grid is slightly inconsistent with logic elsewhere in CISM

[escomp]

Glad's is_in_active_grid is slightly inconsistent with logic elsewhere in CISM

[billsacks]

After changing this, a good sanity check that it is correct is: when running the CISM standard testing (aux_glc tests in the CISM-wrapper test suite), answers should change, but only (initially) in a small number of grid cells: those with topg exactly equal to 0.

One way to confirm that we aren't getting inadvertent changes elsewhere could be: create a temporary input file where grid cells with topg == 0 are replaced with topg = 1. (e.g., ncap2 -s 'where(topg == 0.) topg = 1.;' greenland_4km_epsg3413_c171126.nc greenland_4km_epsg3413_c171126_topgNonZero.nc.) Confirm that this only changes a small number of grid cells in the input file. Then run both the baseline and new code with this modified file (e.g., in test SMS_Ly3.f10_f10_mg37.I1850Clm50SpG.cheyenne_intel), and confirm that they are bit-for-bit. This will demonstrate that any answer changes that arise when using the actual input file come from just this small set of grid cells with topg == 0.

[whlipscomb]

@billsacks – I like your suggestion with the temporary input file.