Factor wrapper interface out of API

[ntessore]

The wrapper interface is a complication that the Cosmology API does not need. It could be moved to a base cosmology.compat package if we found that would be useful; I suspect we will find that it is not useful enough to merit its own package. The same should be true for inclusion in the standard.

We should keep the documentation about wrapping existing cosmology objects, but make it clear how this can be achieved in terms of the individual methods and attributes.

[nstarman]

I'm fine moving it to a companion package.
It does serve as the basis for cosmology.compat.astropy, cosmology.compat.camb, etc, since it tightly defines how an object can serve as a proxy for a cosmology.