Bayesian sample size for exploratory clinical trials incorporating historical data