[PPL-devel] [GIT] ppl/ppl(floating_point): For consistency, use std::numeric_limits<T>: :is_exact in order to check