Ability to pass additional parameters into enriched_mesh_seq constructor from child class of GoalOrientedMeshSeq

[acse-ej321]

I currently have a child class inheriting from :class:~.GoalOrientedMeshSeq where get_initial_conditions, get_solver and other base class functions are dependent on a parameter dictionary. Currently this works well with the methods in associated with :class:~.MeshSeq and :class:~.AdjointMeshSeq but breaks with the get_enriched_mesh_seq method of :class:~.GoalOrientedMeshSeq as it does not allow for passing of any additional parameters which may be associated with the child class.

The initial question I have is, would it make more senses for me to refactor my derived class or would it make sense to refactor get_enriched_mesh_seq to be more flexible to passing additional parameters?

[jwallwork23]

I think it'd probably be fine to change the base get_initial_condition and get_solver methods so that they accept **kwargs and pass those to _get_initial_condition and _get_solver, respectively. Would that be sufficient for your issue?

[ddundo]

Duplicate of #97?

On a semi-side note, I think it might be good to even require the "object-oriented" approach. I think it's very neat. That would also solve #22 nicely

Edit: actually I think the proper way to do this would be to introduce a Solver helper class, which would be separate from MeshSeq. So I opened #238. Enriching the mesh shouldn't affect the solver at all, so I think adding these options to pass **kwargs would be counter-productive, since it's not the underlying issue. I opened #239 with a quick and dirty example of how that would look. If you like it, I'm happy to do it properly :)