Fix `Eq` and `PartialEq` implementation for Boolean network

[daemontus]

At the moment, Eq and PartialEq for Boolean network rely on a specific variable/parameter/regulation ordering. This is reasonable, but it is not exactly intuitive. It would be better to test regardless of ordering (especially for regulations).

[daemontus]

The question is how do you handle equivalence of Boolean functions. We will probably need multiple equivalence methods with different semantics:

• Identity: two networks are exactly the same, including variable/parameter identifiers. This ensures that two networks will have the same symbolic representation for example.
• Syntactic equality: two networks contain the same variables, parameters, regulations, and functions, but their internal representation can be different (i.e. different variable/parameter IDs, or order in which they are stored when said order does not impact the network structure). This is useful if you want to be sure that two networks will "look the same" in an editor.
• Semantic equality: two networks are isomorphic, in the sense that their graphs are isomorphic and their update functions are semantically equivalent. This is much harder to test because it involves isomorphism, but is useful if we want to actually detect duplicate networks within a data set, or something similar.

[daemontus]

For now, closed with 64a02d2d6b1de45a896b0057d65e76e0af4a1aa0. This effectively fixes the syntactic equality issues and leaves the semantic equality as possible further enhancement.