Is the computerized assessment of psychomotor speed more sensitive to cognitive effects of antiepileptic pharmacotherapy than tests with a focus on higher-order cognitive processing? Implications for the choice of sensitive test parameters