Stochastic mixed model sequencing with multiple stations using reinforcement learning and probability quantiles