Model-based clustering of censored data via mixtures of factor analyzers