Gravar-mail: Sample size re-estimation in crossover trials: application to the AIM HY-INFORM study