Paired t-Statistic

When the same users are measured before and after a change, test the differences $d_i = \text{after}_i - \text{before}_i$:

with $s_d$ the sample std of the differences (ddof=1). Implement paired_t(before, after) returning (t, df) with df = n - 1. If every difference is identical, return (inf, df) (or -inf, or 0.0 if all differences are 0).