[Task] Create a diff wofost72_pp module with one nn.module

[SarahAlidoost]

As an example, we want to show if the partitioning module can be replaced with an nn.module, e.g. partitioning_states = partitioning_nn(dvs).

Requirements:

• Add the new class PartitioningNN in a directory ml_models/crop/partitioning.py, use simple nn.module such as nn.ReLU(), implement partitioning conditions (positive and sum to 1), see this pcse notebook / 11 Optimizing partitioning in a PCSE model.ipynb for physical conditions.
• Use wofost72_pp as base, plug the PartitioningNN in either existing partitioning module or the wofost72 module; depends on the implementation and kiosk, see also the example code block below:
• Add docs and a notebook showing end to end hybrid modeling (with training the model)
• See other requirements in [Backlog Item]: Build hybrid modeling framework #11
• See related notebook [Task] Check if parameters of Partioning should be optimizable #73

import torch.nn as nn

class PartitioningNN(nn.Module):
    def __init__(self, ...):
        ...

    def forward(self, dvs):
        return partition_states/rates


class HybridModel(nn.Module):
    def __init__(self, physical_params):
        super().__init__()
        self.partitioning_nn = PartitioningNN(input_size=3)
        # pass self.partitioning_nn to wofost via config or another implementation
        wofost72_config = Configuration(
            CROP=Wofost72,
            SOIL=WaterbalancePP,
            ML=self.partitioning_nn,  # this can be more generic later
            OUTPUT_VARS=...,
        )
        self.engine = Engine(..., config=wofost72_config)
        self.physical_params = nn.ParameterDict(...)

    def forward(self, ...):
        model = self.engine.setup(...)
        model.run_till_terminate()
        results = model.get_output()
    return results