npo entailed phenotypes -> subclasses

[tgbugs]

The ontologically correct approach to entailed phenotypes is to have another cell type that is subClassOf the equivalent class necessary and sufficient condition type that is itself subClassOf the entailed types. This will avoid cases where there are single individual counter examples. In additional we probably want a relationship that we can use to indicate which subclasses we want to use as "high probability entailed phenotype bag" or something like that.