I sympathize, but your new formulation has exactly the same problem as your first one: discernible by whom? If your BAC is 0.0001 but a police officer says that you appeared to be impaired in their subjective judgement, should that be enough to convict you of driving while impaired? And do you even want to try to distinguish alcohol-induced impairment from sleep deprivation? Do you really want to start punishing people for driving while sleepy? What are you going to use as the legal standard for that?
An objective test of reflexes / response time would be highly technically feasible. (I'd also like to see periodic re-certification for drivers with such a test too, but that's another issue.)
So do I. The challenge is coming up with a legal standard to decide who goes to jail and who doesn't which does not result in a ton of false positives. Do you really want to turn going to work after a bad night's sleep into a crime?