Allow for user-configurable physical constants

[spencerkclark]

For planetary atmospheres other than Earth's, physical constants like the gravitational acceleration and the planetary radius will be different. To accommodate results from simulations of these atmospheres, aospy will need a mechanism to allow users to set these constants themselves (rather than have them be hard-coded to Earth's values).

xref: @spencerahill's https://github.com/spencerahill/aospy/pull/223#issuecomment-340331253

[spencerahill]

This is a particular example of the broader issue of supporting user-specified config, c.f. #145. But I think it's worth having this separate issue to track this, as #145 is pretty open ended.

[spencerkclark]

Right, I see #145 as more global configuration-type stuff; this is likely to vary from Run to Run or Model to Model, so it might require a different approach.