Add classmethod to construct Package objects from iMOD5 data

[JoerivanEngelen]

The iMOD5 data model has the following peculiarities we need to deal with:

• Data for a package can be defined on different grids (extent, grid sizes). Most common is grids with different extents.
• Boundary conditions can be defined as planar grids, which need to be distributed across layers.

Good to know:

• Caching is probably less important, as all data is structured
• iMOD5 variables are defined on layers consistently amongst each other within one system

For iMOD Python this requires the following work:

• Regrid all variables to the same grid
• Allocate boundary conditions to layers if defined as planar grids (see: For RIV, DRN and GHB packages it should be possible to automatically assign them to the appropriate layers of the model. #628)
• Each iMOD5 river/drain system needs to become a separate package

Add classmethod to Package which can initiate a new package.
In pseudocode:

class Package:
...
    @classmethod
    def from_imod5_package(cls, dis_pkg, **data):
        # target grid is idomain
        target = dis_pkg.dataset["idomain"]
        # Regrid to one consistent dataset
        dataset = cls._regrid_like(target, **data)
        return cls(**dataset_allocated)
...

For the Structured Discretization, the target grid is the smallest grid in the package.
This is picked up in #964
Therefore the method needs to be overloaded:

class StructuredDiscretization:
...
    @classmethod
    def from_imod5_package(cls, **data):
        # Find most fine-grained gridsize in data to regrid to
        # Clip to smallest extent
        target = find_smallest_grid(data)
        # Regrid to one consistent dataset
        dataset = cls._regrid_like(target, **data)

        # Add logic to prepare the dataset in the following way here:
        # IBOUND < 0 -> IDOMAIN = 0
        # TOP 3D -> TOP 2D
        # Thickness <0 -> IDOMAIN = -1

        return cls(**dataset)

further refinement 06-05

This issue focusses on packages that are in the Flow model, and that are not boundary conditions, and that contain grid -data (unlike the buoyancy package for example). This means sto, npf and ic. But npf was covered in a previous issue, so for this one it's sto and ic. for this issue.