parameter dependant model

Hi,

This is not really an issue. I would like to hear your view on making the pde formulations parameter dependent. I am using your package you my [bifurcation tools](https://github.com/rveltz/PseudoArcLengthContinuation.jl). I have written a simple wrapper to use my "newton".

I guess the model parameters could be put in your `cache` from `_new_nlsolve_cache`.

I would like to hear your opinion about this and potential pitfalls.

Thank you,