Create input schemas; match template model with it

[bdarcus]

Add JSON schemas (though use the YAML representation) for citation and references.

• csl-next-input-citation.yaml
• csl-next-input-citation-reference.yaml
• csl-next-input-references.yaml
• csl-next-input-reference.yaml

Can also combine the first two and last two.

Data types:

• object
  • contributor
  • author
  • editor
  • translator
  • publisher
  • title
  • locator
  • citation-reference
• edtf date (it's a string, but with a special datatype)
  • issued
• string
• list (containing potentially any of the above)
  • citation
  • bibliography
  • contributors
  • author-list (is it feasible to a have list of mixed objects?)
  • locators

Not worrying about for now: rich text, which would at least optionally make all strings lists of objects.

For contributor vs agent vs publisher, see Dublin Core. In the BIBO RDF ontology, which I co-authored, publisher is a relation; an "ObjectProperty", that derives from dcterms:contributor, so I've followed that above.

An agent is an RDF class in that world, independent of the relation between it and a work it contributed to. We don't need to get too pure though.

This is a generalization of #9.

[bdarcus]

@denismaier - my plan is to tackle this next, sometime in the next few days to week, and then revise the template model to match it.

I think at that point it should be ready to discuss more widely.