It'd be really useful to have a collection of image search results with their position in the search results and their elasticsearch scores, where we can rate the images as "good", "bad", or "indifferent". It'd allow us to see which of the search signals are strong, and which are weak

There are two steps to this

1. gather search results for individual search signals (e.g. statement, text, file_text, etc) and store them
2. provide an interface so that people can rate an image as a good (or otherwise) result for a particular search term