Reproducibility of the WHO classification of thymomas: practical implications