Implement compute_full_pmodel_outputs and unit tests

[ychnli]

This is a duplicated PR of #1198 because git history complications made it very cumbersome to rebase. See #1198 for previous discussion. The description is copied here (and updated with latest names) for convenience:

This implements of the core P-model computations in the function compute_full_pmodel_outputs (#1197).

To-do

• still need to write update_photosynthesis! in pmodel.jl with appropriate subdaily timestepping equations
• still need to make sure that stomatal conductance is appropriately updated, since P-model predicts it jointly from the optimality criteria

Content

• compute_full_pmodel_outputs and associated helper functions in pmodel.jl and canopy_parameterizations.jl
• test_pmodel.jl unit tests and test files on ClimaArtifacts

Important note

in this PR, the PModel cannot be used yet in the canopy model because subdaily GPP variability is not explicitly separated from longer timescale acclimatized GPP. Instead compute_full_pmodel_outputs should be used as a ClimaLand implementation of the rpmodel package that models GPP variability assuming adjustment to optimality on weekly timescales.

[juliasloan25]

I just took a brief look - I'll look more at the src changes after our meeting today.

[kmdeck]

Are the units correct here? po2 is a Pressure, but I think Ko and Kc are in units of mol/mol

[ychnli]

Kc and Ko are Michaelis-Menten parameters that have the same units as substrate concentration (ref: Wikipedia). Here the substrates are gases, so the units of Kc and Ko can either be mol/mol (mixing ratio) or Pa (partial pressure). Since Kc and Ko are scaled via Arrhenius functions from their values at standard conditions (25 C, 1 atm), the Pa units vs. mol/mol units differ by a constant factor of 101325 Pa. The constants Kc25 and Ko25 I use are in pressure units and are the same as Stocker (2020)

I can add a comment detailing this in the code, or maybe we have a preference as to what should be standard. Currently the docstring description for the parameters is wrong, so I'll fix that.

It seems like the Farquhar code calculates everything in mol/mol, without converting to partial pressure. But as long as we are consistent, the constant factor of $P_{atm}$ cancels out in the end when we calculate $A_c$:

$$A_c = V_{cmax} \dfrac{[CO_2]_{i} - \Gamma^* }{[CO_2]_i + K_c(1 + [O_2]_i / K_o)}$$

(the numerator and denominator are both concentrations, so whether we use mol/mol or Pa cancels out)

[kmdeck]

I'm good with whichever approach, as long as we are consistent like you wrote, so the factor cancels out in the end, and as long as we are consistent between the two formulations

[ychnli]

I added a comment detailing this when we declare the PModelConstants struct.