Finding Counterfactually Optimal Action Sequences in Continuous State Spaces