Better control over nodes used for the mapping.

[VEZY]

We need to be able to control which nodes in the MTG are used to map variables for a model. For example, at the time if we want to have a model that allocates carbon preferentially to the leaves closer to it, we could do this:

mapping = Dict(
    "Phytomer" => (
        MultiScaleModel(
            model=SpatializedCAllocationModel([parameters goes here...]),
            mapping=[
                :carbon_assimilation => ["Leaf"],
                :carbon_demand => ["Leaf", "Internode"],
                :carbon_allocation => ["Leaf", "Internode"]
            ],
        ),
    ),
    "Internode" => (
        CDemandModel(...),
    ),
    "Leaf" => (
        PhotosynthesisModel(...),
        CDemandModel(...),
    ),
)

But atm, PlantSimEngine would provide the values of all leaves in the plant, and if there are several plants, it would provide all leaves of all plants.

What we should do by default here is to provide only the leaves from the phytomer. What we should do:

To map a lower scale to a higher scale, we should only provide the lower scale that is "inside" the higher scale. For a mapping of a higher scale to a lower scale, we could search for all higher scales if it is given as a vector (e.g. ["Plant"]), or the first one if it is given as a scalar (e.g. "Plant").

Of course we should also consider how the package manages adding new organs, as it will change such mappings, but not for all organs.

Note: we can already manage such simulations using the node provided by default as input to the model (when #75 is merged).