add get_ligand_network helper functionality to AlchemicalNetwork

[atravitz]

We currently have ligand_network.to_rbfe_alchemical_network() functionality, but no helper for extracting the underlying ligand network from an alchemical network.

This is currently done in some of our downstream scripts as

   edges = []
    for e in alchemical_network.edges:
        edges.append(e.mapping)

    network = LigandNetwork(edges=set(edges))

but it might be useful to add this as an AlchemicalNetwork method.

[dotsdl]

I don't believe it's always the case that an AlchemicalNetwork can translate to a LigandNetwork, as not all AlchemicalNetworks encode RBFEs between ligands. For example, it's possible to have an AlchemicalNetwork that encodes ABFEs, or protein mutations, or all of the above.

It's not clear to me what the expected behavior of this method should be given this.

[IAlibay]

Echoing @dotsdl's comment, this is a bit of a messy situation. LigandNetworks are a convenience for creating AlchemicalNetworks, but they aren't really built to roundtrip to/from AlchemicalNetworks. Indeed as @dotsdl mentions, they aren't really needed for AFEs.

The above code would just return you an empty network for an AFE because all mappings are None.

Ideally we would need to get a set of ligand nodes which we could then add to the network, but we would have to assume that the alchemical species is the ligand (which means assuming Protocol behaviour).