Frac_sno should generally be 1 over lakes and wetlands if there is any snow

From a discussion with @dlawrenncar and @swensosc - Scientifically, frac_sno should probably generally be 1 (or close to it) over lakes and wetlands if there is any snow: Since these landunit types don't have any topography, we would expect uniform snow cover when there is any snow.

For the melt piece of frac_sno, this can be achieved with appropriate choice of n_melt in this code:

https://github.com/ESCOMP/CTSM/blob/daf1459fea6c3191f4218242c98a5ebf7bebe846/src/biogeophys/SnowCoverFractionSwensonLawrence2012Mod.F90#L454-L466

That doesn't deal with the change in frac_sno during accumulation, but @swensosc says that frac_sno goes pretty quickly to 1 during accumulation so that's not as important to change.