Some data on surface datasets can be in single precision

[ekluzek]

In the new ctsm5.2.0 datasets we brought the new soils data in as single-precision verses double precision which had been the standard for everything on the datasets in ctsm5.1. Since they've been used in simulations and also to make all the supported resolutions, we consider this well tested.

Anything that is both summed to 100_r8 in mksurfdata and CTSM -- does need to be in double precision. This includes almost all of the PCT_* variables. PCT_CLAY and PCT_SAND don't need to follow this because PCT_SILT is ONLY calculated inside of CTSM the residual between 100_r8 and the sum of the other two.

So there are a lot of variables on surface datasets that could be made float rather than double.

In ctsm5.2.0 here are the list of variables coming in as single precision:

	float PCT_SAND(nlevsoi, lsmlat, lsmlon) ;
	float PCT_CLAY(nlevsoi, lsmlat, lsmlon) ;
	float ORGANIC(nlevsoi, lsmlat, lsmlon) ;
	float ORGC(nlevsoi, lsmlat, lsmlon) ;
	float BULK(nlevsoi, lsmlat, lsmlon) ;
	float CFRAG(nlevsoi, lsmlat, lsmlon) ;
	float PHAQ(nlevsoi, lsmlat, lsmlon) ;
	float SLOPE(lsmlat, lsmlon) ;
	float STD_ELEV(lsmlat, lsmlon) ;