The Theoretical Reward Learning Research Agenda: Introduction and Motivation — AI Alignment Forum