As a learning exercise, use ElasticSearch to create an index of the benchmark data

• Index some subset of the BioCADDIE data
• Run some queries
• Figure out how to change the similarity module

We'll probably use LM Dirichlet in the end