Methods for including information from multi‐arm trials in pairwise meta‐analysis