Optimal quantile hedging under Markov regime switching