Run the expertise model for one group of users against another group

[haroldrubio]

For some cases, we will want to compute scores for one group against another (i.e area chairs vs senior area chairs). In order to do so, fetch the publications for 1 group and place them in submissions.json. For the other group, place their publications in an archives folder. Lastly, perform some averaging across the raw scores from the expertise model. Also, add this functionality under a new expertise API endpoint.