Subclass sync is underreporting "support"

[matentzn]

We have had a lengthy discussion about whether it is better to generate supporting evidence against mondo.owl or mondo-edit.obo, and have determined that there is some advantage of using mondo.owl, hence we closed #568.

This supports evidence gathering against inferred axioms:

mondo-edit.obo:

X sub Y
A = D and R some X
B = D and R some Y

mondo.owl:

X sub Y
A = D and R some X
B = D and R some Y
A sub B # <---- if an ontology we are sourcing from supports this, great, we now get that evidence in.

However, this does not take into account transitive reduction. Let us say we have this case in the mondo-edit file:

mondo-edit.obo:

A sub B
B sub C
A sub C # <-- Redundant, because C is a grandparent of A.

mondo.owl (redundant axiom removed):

A sub B
B sub C

If an external source supports A sub C, we do get that information attached to the edit file. This means, curators are missing some potentially interesting information (namely that A sub C is supported by source Z) when making modelling decisions.

I suggest to revisit #568 and merge this with what we have now, taking both mondo.db (based on mondo.owl) and mondo-edit.db (based on mondo-edit.obo) into account when collecting subclass axiom evidence.