Split comparative models into their constituents and annotate

[benmwebb]

Some of the comparative models for subunits are misleading in that they are really multiple models that have been manually concatenated. For example, Nup84 is actually two comparative models, one for the N terminus and the other for the C terminus. These should be split up into the original models, so that each file can be accurately annotated (by combining two models into one, it isn't clear which annotations, such as model score or templates used, apply to which submodel).