sometimes get NaNs from GSIC

[bvegawe]

GSIC produces NaNs if it all melts.

[bvegawe]

Perhaps this isn't really an issue? In my calibrations of the Hector model, sometimes I get NaNs when technically it should be V0, but bugging out is probably not a horrible response either (I presume full melt is fairly unrealistic under normal circumstances).