Suitability of calibration polynomials for eye-tracking data with simulated fixation inaccuracies.