freezing of rain onto shallow snow layers

[har917]

CABLE (recently modified by PR #527) allows for the freezing of rain onto a snow layer (and/or frozen soil) - with energy/temperature adjustments accordingly.

The sequence of calls effectively allows

1. all rain to be frozen - even if this pushes the snow temperature above freezing (cbl_snowAccum)
2. new snow layer to be remelted (second call to cbl_melt) while keeping ssoil/snow layer temperature at/above freezing.

This is physically inaccurate - what happens in reality is that only the fraction of rain necessary to raise the temperature to zero would be frozen. We would be better placing limit on the amount of rain that can be frozen while keeping the surface temperature at/below freezing.

Additionally - dependent on the precise sequence of when %gammzz is evaluated - this split process risks breaking energy conservation.