multiple testing, FDR

[pbstark]

• Implement multiplicity adjustments: Bonferroni, Bonferroni-Holm, Westphall-Young.

• Also Sidak? Or should we omit methods that require independence or positive regression dependence?

• Implement one or more FDR adjustment methods.

• write a "wrapper" for multiplicity adjustment with different methods as parameters, or call different top-level functions?

Westphall-Young is close in structure to NPC: need to track individual permutations).

FDR, Bonferroni, Bonferroni-Holm operate on P-values.