How well do self-interaction corrections repair the overestimation of static polarizabilities in density functional calculations?