prepraster() should work differently in the absence of elevation in predictors

[courtiol]

prepraster() takes isofit as input, so if the model does not contain elevation as a predictor, it should drop the values.