Implement different approximations for the metric terms on the sphere

[tristanmontoya]

In the Cartesian approach for the spherical shell, we should be using the invariant curl form metrics, as implemented here. It remains to be determined if this is the best option for the covariant form, and it might be best to have the option to select between different approaches for computing the metric terms.

[tristanmontoya]

Andrés has implemented a nice approach to this in #36.