Kappa eta fix

[dihm]

This fix ensures Cell.kappa and Cell.eta use a dipole-allowed edge of the coupling to introspect q

Fixes #45

[dihm]

@benjamin-n-miller This could really use your eyes. I'm having a hard time believing the way I implemented this is really the best way.

[Json-To-String]

It seems like you were right that the logic also existed in the coupling coefficient matrix:

rydiqule/src/rydiqule/sensor_solution.py

Lines 417 to 433 in a256cfd

	#integer values of each state that is in the upper and lower manifolds
	states1 = sensor_utils.match_states(coupling[0], list(self.couplings.nodes))
	states2 = sensor_utils.match_states(coupling[1], list(self.couplings.nodes))
	# have to rebuild graph only with edges we want
	# because networkx defaults to 1, not zero
	# on adjacency creation
	ag: nx.DiGraph = nx.DiGraph()
	ag.add_nodes_from(self.couplings.nodes)
	sg = self.couplings.subgraph(states1+states2)
	se = tuple(edge
	for edge in sg.edges(data='coherent_cc')
	if edge[2] is not None)
	ag.add_weighted_edges_from(se)
	coupling_adjacency_matrix = nx.to_numpy_array(ag).T.conj()
	return coupling_adjacency_matrix

with some slight differences