On The Evidence of Testimony For Miracles: A Bayesian Interpretation of David Hume'S Analysis