Towards a typology of gender‐related school effects: some new perspectives on a familiar problem