Gravar-mail: Efficient estimation of stereo thresholds: What slope should be assumed for the psychometric function?