feature request: allow specifying contrasts and p-value correction in posthoc()

[yanxianl]

Hi,

The recently added features to allow comparisons for more than 2 groups with post hoc tests are just great! However, pairwise comparisons are not always what we need. Instead, we may be interested in particular subsets of contrasts. It'd be great to allow for customized contrasts and the p-value correction method in the posthoc function.

-Yanxian