On an unbalanced growth curve model with random effects and AR(1) errors from a Bayesian and the ML points of view