Rigorous definition of symmetric and asymmetric influence diagrams

[jaantollander]

We need a rigorous definition of symmetric influence diagrams. Influence diagrams which are not symmetric are asymmetric. If an influence diagram has inactive chance states, it can be asymmetric, but I don't think it's a sufficient condition.

We can only use activate_path_cut on symmetric influence diagrams where the exact number of active paths is constant.