You've misunderstood the theorem. It doesn't say you can't achieve any combination of mostly C/A/P; it says you can't have perfect C, A, and P at the same time.