test different possible materialized view updates to accomodate post-composed phenotypes

[bellerbrock]

From Tuesday's discussion - the plan is, using topic/postcomposing branches, to load demo post-composed data in the fixture. The demo data doesn't need to be complicated, terms to compose can come from the cassava trait ontology and by loading simple time and unit ontologies.

Then, using the demo data, test out the possible solutions to storing and accessing post-composed phenotypes in the materialized views. These include storing them with a single group_id, storing the multiple cvterm_ids within a json string, or storing individual rows for each separate term.

[bellerbrock]

After looking at some demo data today, we've come up with a plan going forward for the materialized views and the wizard. The plan is, instead of jsonb arrays, to a add a trait_group view, and a group_id column to the phenoview. And a new trait_components category in the wizard and binary views.