'Ultra-deductivism', perfect knowledge and the methodology of economics: some lingering doubts