What Have We Learned about Attention from Multiple-Object Tracking (and Vice Versa)?