visualize and compare.fits to handle inhibited variable operations? [Enhancement]

[mlotinga]

Great package, especially finding visualize and compare.fit functions so useful for mixed models. I'd love to see these functions able to handle I() variable operations, where the original variable does not necessarily also appear in the original model formula. One example is a reciprocal variable. This kind of use currently raises an 'object not found' error. Of course, this could be added as a new variable to the data frame, but for rapid model exploratory development, it would be good to be able to avoid adding variables (plus, the way I() variables appear in results summaries can be helpful).