An Application of Markov Reward Processes to the Italian Bonus-Malus System