Causal Specification Refactor?

[f-allian]

Is the Causal Specification data class necessary? In all examples we have, the variable causal_specification = CausalSpecification(scenario, causal_dag) is created but never used. Other than for formal definitions (and maybe our unit tests), do we need it or should we remove it?

[christopher-wild]

It appears that the Causal Specification was made as a data class to match the abstract idea of a 'causal_specification'. It was also a useful packing of the scenario and causal_dag for the CausalTestEngine, which had far too many attributes.

Now we don't have an engine I don't think it's as necessary from purely a code perspective. Not sure on how useful it is for it to match the phrasing of the paper though.

It looks like it is still used in the json front end & test suite. But this is just packing the two input classes then immediately unpacking them by the looks of it. So should be not too much to undo that and remove the class!