Large sample inference for a win ratio analysis of a composite outcome based on prioritized components | Synapse