Adding support for base states

[ali-ramadhan]

Not sure how important this is given that we're not super worried about truncation or rounding errors in the vertical momentum equation right now. From my understanding, the errors are more severe when running with terrain-following coordinates so perhaps we don't even base states to do most of what we want to do.

A big reason to add base states though is to provide hydrostatic states in which we can initialize simulations without worrying about transient acoustic waves (see issue #42).

It would also reduce the amount of boilerplate code by a lot. Currently the verification experiments spent a non-trivial amount of code setting up a dry adiabatic atmosphere and running it to steady state. Users cannot be expected to do this.

[ali-ramadhan]

I should mention that the amount of refactoring is pretty small if we add base states (and evolve the perturbations).