Add decompositional approach for IMEX multiphysics systems

This current PR generates the simulation by taking as input two Decapodes, which share identical (by name) tangent variables. This is fine for the purposes of a proof of concept, but one can imagine a more principled approach, using perhaps a UWD with implicit and explicit components, and which share/ expose certain variables. A collage-like approach could have some advantages, but that requires some brainstorming.

One could roughly imagine the sort of analysis that can build off of this is a principled/ “automatic” means of splitting a single Decapode into an implicit and explicit Decapode, but are connected by that span-like structure, and we can verify the decomposition is valid by (monic, natural) homomorphism search from the apex to the original, etc etc

_Originally posted by @lukem12345 in https://github.com/AlgebraicJulia/Decapodes.jl/issues/369#issuecomment-4320494532_