Comment on "Estimating mixtures of normal distributions and switching regressions" by Quandt and Ramsey