This is right and wrong: AFNI's 3dClustStim did contain a specific bug (since corrected) and is used for fMRI analysis, but the problem that the authors identified is more general than that and encompasses all of the tested packages.