[Epic 1]: Optimizing parameters in Wofost72_PP using pytorch

The goal is to make the end-to-end wofost72_pp differentiable, allowing gradients to flow through the simulations for optimization. See Models available in PCSE. The current structure of PCSE regarding Wofost72_PP:

• class Wofost72_PP(Engine)
  • └── class Engine(BaseEngine)
    • └── class Wofost72(SimulationObject)
      • └── 9 modules
    • └── repository WOFOST_crop_parameters: branch wofost72

Milestone:

Demo a prototype of optimizing parameters = ["TDWI", "SPAN"] in differentiable wofost72_pp, see a similar notebook using PCSE.
Note: the module Root_Dynamics contains "TDWI" and the module Leaf_Dynamics contains "TDWI" and "SPAN".

Backlog items

Start with Wofost72_PP in pcse/models.py, Engine in pcse/engine.py, and Wofost72 plus its modules in pcse/crop/wofost72.py; inspect Root_Dynamics and Leaf_Dynamics for TDWI and SPAN. Review the linked parameter-optimization notebook and confirm done with a prototype that optimizes TDWI and SPAN while gradients flow through the end-to-end wofost72_pp simulation.