Why `getManifolds(vd::VariableNodeData)`

[dehann]

Should variableType not just exist at DFGVariable level? Why all the way down to VND...

[dehann]

@Affie , do you perhaps remember why we made variableType a VariableNodeData level thing? My expectation was/is that each variable can only have one variableType. solveKey (super-solving) is for numerical differences only, not for mixing different variableTypes.

[dehann]

Going further, why did we have VND{T <: InferenceVariable}, and not just DFGVariable{T}? This is more of a DFG consideration.

[Affie]

It might just be historic since softtype used to be a field in vnd. With the replacement variabletype as a singleton type, there shouldn't be a reason to store the type itself.
I can't see a reason it has to remain as a field. I don't know if the type is needed somewhrere though.