[PPL-devel] [GIT] ppl/ppl(floating_point): Use ROUND_UP (or, in one case where intervals are involved, ROUND_DOWN)