Correct the minor formula mistakes in `gs_info_rd`

[LittleBeannie]

In the paper "Minimum risk weights for comparing treatments in stratified binomial trials",

• delta is defined by p_e - p_c, rather than p_c - p_e

• f_i is the prevalence ratio, which should be n / sum(n) rather than n/max(n)