Evaluation of bi-directional causal association between depression and cardiovascular diseases: a Mendelian randomization study