How to compute weighted standard deviation in dplyr