Remove self-subclasses

[hina-shah]

To reproduce: In question builder, try submitting a query "asthma" relates to "asthma". This will create a result table, and one of the rows should be from asthma to asthma. If you click on the row it looks like here in the image:

[cbizon]

I think we should filter out A - subclass-of - A edges somewhere. Logically true; totally useless.

[EvanDietzMorris]

Is this something we could apply more generally? Are there any A-->A edges in robokop we care about? The issue is complicated by the fact that this can occur during normalization.

[cbizon]

I think that there are probably self-interaction or self-regulation edges that we would want to keep

[EvanDietzMorris]

cool, I think if we want zero A-subclass_of-A edges we can pretty easily apply that in normalization by default, but it should probably be it's own operation that can be specified in the graph spec