Additional changes needed during normalization process

[MarkDWilliams]

Ref: NCATSTranslator/Feedback#134
During the course of changing nodes in results and knowledge_graph as part of the node normalization process, we neglected to apply these same changes to subject and object fields on edges. These changes should be extended to include the CURIEs which exist as values for subject and object on KG edges as well.

[cbizon]

FWIW, there is a (badly-named) endpoint on NN /response that normalizes an entire TRAPi response. It normalizes the nodes as well as the edges, and it supposedly handles the case where e.g. two nodes on the input KG normalize into a single node on output. I say supposedly b/c I don't think it has ever gotten used widely, but you may find it useful either from calling or from looking at its implmentation.

[MarkDWilliams]

Good looking out. I did not recall that and will look into it.