Depth-dependent stability functions for surface fluxes

[cbegeman]

This PR adds additional options for surface fluxes, namely:

• "quadratic" drag law that does not consider stability (law of the wall)
• linear stability function for drag (linear in M-O length scaled depth) first proposed by Businger et al. (1971)
• depth-dependent formulation for scalar fluxes following McPhee et al. (1987) and implementation of Ramudu et al. (2018)

The main namelist parameters controlling surface flux parameterizations are

• most_method characterizes method of solving for surface fluxes
• drag_law characterizes functional form that determines momentum fluxes (e.g., 'quadratic' to neglect stability, 'businger' to choose linear stability function and 'rotation' to choose method from McPhee (1981))
• gamma_mcphee which, for the Mcphee MOST method, determines whether scalar fluxes should be determined assuming the mixing length is affected by proximity to the boundary (gamma_mcphee = 'depth-dependent', Ramudu et al.'s (2018) implementation) or is not affected by proximity to the boundary (gamma_mcphee = 'constant', Holland and Jenkins (1999) and most ocean models' implementation).

[cbegeman]

@xylar This is ready for your review. Our default option will likely be most_method='mcphee', drag_law='businger', gamma_z_dependent=True. The 'businger' modification to scalar fluxes further reduces the turbulent component.

[cbegeman]

@xylar Thanks for looking this over. I moved the stability check outside of loops as you suggested.