Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders