Merging models take 2

[SouthEndMusic]

A few moons ago I made an attempt at supporting merging models in Ribasim Python, which was never merged:
Issue: #912
PR: #914

A few things have changed in Ribasim Python since then, most notably:

• having non-unique node IDs, which we must use here,
• the Add-API, which I suggest we do not use, and directly merge dataframes just as in the previous attempt.

New merge method structure

I propose the following structure for a method model.merge(other):

• First obtain the highest subnetwork_id in model:

subnetwork_id_shift = self.max_subnetwork_id()

where self.max_subnetwork_id is a method which loops over the now distributed chunks of the node table to find the maximum subnetwork ID;

• Now loop over all node types (e.g. node_type = basin) and call a merge method on the level of NodeModel (?):

node_id_type_shift = self.node_type.merge(other.node_type, subnetwork_id_shift)

This merge method looks something like this:

function merge(self, other, subnetwork_id_shift) -> Int:
   node_id_shift = max(self.node.df.node_id)

   # merge node table chunks using the node_id and subnetwork_id shifts
   ...

   # merge other tables using the node_id shift
   ...

   return node_id_shift

• Now the edge tables can be merged using the node ID shifts from all different node types. What to do with the edge ID?
• Special care has to be taken with tables from node types which refer to nodes from other types (i.e. control node types), in which the node ID shifts also have to be applied.