Add more subq patterns during Phenio transform

[caufieldjh]

Phenio is relaxed by robot during the transform step, but this doesn't change every equivalence axiom to SubClassOf, just the "first level" (i.e., it doesn't act recursively).
See also ontodev/robot#1007

There is likely a more complicated way to so this with SPARQL alone by defining what a nested subq pattern looks like, or we can just wait for it to be implemented in robot.

[matentzn]

This PR uses ROBOT to achieve the same thing w/o the recursive method:

mgijax/mammalian-phenotype-ontology#3480

Even if the recursive method works, it wont be enough for phenio, because you want the direct relations across the haspart buffer. It would really be good if you could just comment on the above PR saying that it would be great if that could be implemented soon, because its hard for me to push it through..