On the role of diversity measures for multi-objective test case selection