On auroral electrojet curvature and the interpretation of azimuth-scan, radar Doppler data