[Dycore] Acoustic Substepping W/ Moisture

[AMLattanzi]

Summary

This PR explores formulations for the acoustic sub-stepping algorithm with moisture. The approach taken here is to not transport theta_m but to construct theta_m by including q_v as a fast variable.

Approach

Additional Changes

Couple the acoustic solver with macrophysics. Tendencies computed for $$\theta_d; q_v; q_c$$

Todo

• Protect against the case of MoistureType::None
• Fully migrate qv update to slow pre (diffusion)
• Implicit diffusion of qv as a fast variable
• Correct the _NS sub-stepping file

[AMLattanzi]

Running WPS for 10 steps with development and this PR:

Dev: Cloud fraction 0.258 (oscillated in time)
PR : Cloud fraction 0.258 (oscillated in time; coupling = false)
PR : Cloud fraction 0.941 (monotonically grew in time; coupling = true)

WPS inputs changes:

erf.fixed_dt  = 10.0 
erf.fixed_mri_dt_ratio = 6

erf.macrophysics_acoustic_coupling = true/false

erf.dycore_horiz_adv_type    = Upwind_3rd
erf.dycore_vert_adv_type     = Upwind_3rd
erf.dryscal_horiz_adv_type   = Upwind_3rd_SL
erf.dryscal_vert_adv_type    = Upwind_3rd_SL
erf.moistscal_horiz_adv_type = Upwind_3rd_SL
erf.moistscal_vert_adv_type  = Upwind_3rd_SL

erf.implicit_before_substep     = true
erf.implicit_thermal_diffusion  = true
erf.implicit_moisture_diffusion = true
erf.implicit_momentum_diffusion = true
erf.vert_implicit_fac           = 1. 1. 0.

[AMLattanzi]

The vertical advection here often leads to negative $$q_v$$. The swing in $$\Omega$$ between substeps in this PR does not appear to be substantially larger than the development branch either.