Epistemic uncertainties and natural hazard risk assessment – Part 2: What should constitute good practice?