Make better use of crossref data
Crossref returns a lot of quite useful information (publisher, licence, event the document is from) that we can make better use of for storage.
At the moment, I'm storing the minimum that I needed for testing - it would be better to store all/most of the relevant information.