The (Mis)Specification of Discrete Time Duration Models with Unobserved Heterogeneity: a Monte Carlo study