Design schema for pre-computed owlsim experimental data

We need a schema to be applied to the owlsim experimental datasets to be loaded into disco. The data is pair-wise comparison data (such as comparing the similarity of all diseases to each other). The results of each comparison are a list of scores and comparisons themselves.

The schema design will also go into creating new serializers for owlsim output.

based on discussion at the hackathon, it might look something like:

3 (or 4) tables:
1. experimental metadata (to be specified in another ticket)
2. entity-x-entity, with values being an index into another table
3. comparison index and the results. this data includes pairwise-comparisons of ontology classes, including least-common subsumers, similarity scores of entity-comparison per metric. it is possible the the pairwise-comparison of classes should be yet another table to reduce redundancy.