Investigating the Benefits of Nonlinear Action Maps in Data-Driven Teleoperation