Write out file with semantic similarity of all pairs of HP terms

Workflow:

• download HPOA file in order to calculate frequencies for each term
• download HPO
• turn HPO into DAG
  • drop singletons
  • drop non-HP terms
  • confirm DAG-ness
• load graph into Ensmallen
• make Resnik model, calculate all by all Resnik similarity
• make Jaccard model, calculate all by all Jaccard similarity
• pandas join tables
• write out to file
• upload file to S3 (KG-Hub?) # decide on this later