Pyrealm

- Addition of a parallel C4 method for the `lavergne20` CalcOptimalChi method. The
methods are now called `lavergne20_c3` and `lavergne20_c4`.
- Addition of default theta model parameters for `lavergne20_c4` giving beta predictions
as 1/9 of those for C3.
- Update of soil moisture option handling in PModel to avoid conflicting approaches
(rootzonestress, soilmstress, lavergne20_cX).
- Updated docs for the CalcOptimalChi methods and soil moisture page.
- Addition of an explicit ExperimentalFeatureWarning - currently rootzonestress and
lavergne20_c4.

- Implementation of alternative methods for CalcOptimalChi, including Lavergne et al
2020 soil theta estimation of beta, c4 with negigible photorespiration.
- Addition of optional soil theta to PModelEnvironment, underpinning the lavergne2020
CalcOptimalChi method.
- Alteration of PModel arguments. Since there are now different options for simulating
C3/C4, the c4 argument is replaced with method_optchi, which sets C3/C4 status
internally from the method selected.
- Refactor and integration of Alienor's CalcCarbonIsotopes and C3C4Competition models,
from:
[https://github.com/Alielav/pyrealm/tree/alienorlavergne](https://github.com/Alielav/pyrealm/tree/alienorlavergne)
- Refactor of utilities TemporalInterpolator and DailyRepresentativeValues to handle
multiple dimensions and ragged arrays of indices.
- Extended pytest framework to include TemporalInterpolator, DailyRepresentativeValues,
CalcCarbonIsotopes and C3C4Competition.

- Breaking change to inputs to CalcOptimalChi - now uses PModelEnvironment object
directly, not named args for kmm etc.
- PModel testing update: new set of input test values of 100 values across envt space,
not just 4 arbitrary values. Updated R outputs.
- Restructure of test_pmodel - much cleaner use of parameterisations args.
- Addition of units to utilities.summarize_attr and extensive addition of units
throughout docs.
- Flexibility in the units of PPFD - previous versions were agnostic about However, that
leads to nonsensical values of Jmax and Vcmax, which must the units of PPFD - so that
the scaling of GPP could be set by the user. be in µmol m-2 s-1. So, PPFD now must
also be in µmol m-2 s-1.
- Bug in Jmax calculation - not carrying ftemp_kphio correctly into calculation -
corrected by baking ftemp_kphio correction to kphio early.
- Breaking change to calculation of Jmax and Vcmax. Previous versions followed rpmodel
in using a more complex calculation for Jmax and Vcmax that allowed Stocker's
empirical soil moisture effects (beta(theta)) to be worked back into Vcmax then Jmax,
rd and gs. pyrealm no longer does this: the soil moisture correction is applied _only_
to LUE and the getter function for vcmax, jmax, gs and rd issue a warning that they
are uncorrected when soil moisture effects are applied.
- Internal changes - the CalcLUEVcmax class has been retired. This structuring was
integral to the soil moisture correction approach, but with that change, a simple
JmaxLimitation class replaces it.
- Updated value of param_classes.PModelParam.soilmstress_b to published default

- Bugs in calculation of Jmax and g_s fixed.
- Fixed issue with utilities.summarize_attr with masked arrays containing all NaN values

- Fixing the calculation of stomatal conductance for C4 plants - not infinite
- Added estimate_productivity.md in docs to show behaviour of those variables for C3 and
C4 - revealed some issues!

- Updated CalcOptimalChi to return an actual estimate of chi for C4 plants, not just
1.0. Updated documentation and examples to illustrate.