Could/should this index be populated directly from S3 results at time of running of gather_deed_images + gather_image_hits management commands? Should database be skipped altogether? If so, how would we manually mark things like "don't actually upload this one" (i.e. manual override of match/exemption status)