Develop initial demo of tool for anomaly detection

[pleary]

This first pass was implemented as:

• A method in the inatVisionAPI that can process geo scores for thousands of observations with taxon_id/lat/lng very quickly, turning them into a ratio of geo_score/taxon threshold
• A rails model to capture geo_scores and store them in the DB so the geo_score API doesn't need to be called every time an observation is updated (e.g. description is modified, quality grade is change)
• A script to loop through all observations, sending them to the geo_score API, recording the scores in the DB, and updating the observation documents in Elasticsearch w/o needing full reindexing
  • What is not done is any processing of new or modified observations, e.g. when location or observation taxon changes
• Adding geo_score as a sort option to the API
• Adding geo_score as a sort option to the Identify UI