New evaporation scheme

[kmdeck]

Purpose

• Implement evaporation scheme of Lehmann et al, Equation 10: https://agupubs.onlinelibrary.wiley.com/doi/pdfdirect/10.1029/2018GL078803 when q_sfc > q_air.
• Compute sublimation with a separate call to SurfaceFluxes, use a beta factor to ensure this goes to zero when ice content goes to zero

Large changes to src/standalone/Soil/energy_hydrology.jl, all else is updating scripts and tests

To-do

Content

Changes evaporation scheme by writing a new turbulent_fluxes method for EnergyHydrology.
Computes ice and liquid water fluxes separately, so ice_frac is no longer needed, and we now compute vapor_flux_ice
and vapor_flux_liq and store them separately.
Remove/comment out code from previous scheme (surface_resistance, dry_soil_thickness, tortuosity etc functions)
Update scripts/tests

---

• I have read and checked the items on the review checklist.

[AlexisRenchon]

Thank you for the big PR! It looks great!

I bring two things up in this review:

• do we need to add vapor_flux_ice to diagnostics in this PR?
• going through this, I realized that it would be helpful to make a table in the docs containing as many acronyms as we can, with additional info ("stands for", "units", "comment"), somewhat akin to the diagnostics table, but with all sort of acronyms that we use throughout ClimaLand code. This table would probably be huge but very useful, and users would find what they look for via ctrl+f. Examples: Y, p, t, θ_i_sfc, z_0m, ts_in, etc. etc. We could refer to this table in many places.

(obviously this should be in another PR)

[AlexisRenchon]

You mean add vapor_flux_ice?
But it currently doesn't exist?

[kmdeck]

Yes, in this PR we added it. eventually it would be nice to have sublimation as a diagnostic also, and I could add it in this PR?

[kmdeck]

For benchmarking - this PR:

[ Info: Timestep: 900.0 s
[ Info: Duration: 21600.0 s
[ Info: Starting profiling
[ Info: Num samples: 71
[ Info: Average time: 6.68 s
[ Info: Max time: 7.98 s
[ Info: Min time: 6.17 s
[ Info: Standard deviation time: 0.26 s

Previous main:

[ Info: Resolution: (101, 15)
[ Info: Timestep: 900.0 s
[ Info: Duration: 21600.0 s
[ Info: Starting profiling
[ Info: Num samples: 63
[ Info: Average time: 5.55 s
[ Info: Max time: 9.68 s
[ Info: Min time: 4.62 s
[ Info: Standard deviation time: 2.05 s
[ Info: Done profiling

I think this is not surprising given that we now call surface fluxes 2x per step for the soil model, rather than 1x.