Debiased Machine Learning for Conformal Prediction of Counterfactual Outcomes Under Runtime Confounding