It the standards are vague enough to allow some manager somewhere to claim that the CAT II bench meter is good enough for the scenario I laid out, then they aren't helping much. But I'm a bit doubtful that they would be that flawed or that Fluke would deliberately (or even negligently) publish false information that clearly contradicts what you are claiming is the standard. My best guess, based only on speculation and language from Fluke's reference, is that there is additional language specifying a single-phase receptacle/plug, not just any receptacle/plug.
You're on point here, the standards do not have a string of complex clauses for the definitions of Measurement Categories, its laid out plainly as I provided direct word for word quotes of.
Yes, the delineations are not perfect but there is no perfect delineation as the world is complex. This is layers of abstraction as the Measurement Categories define the limits of the safety tests, so all that compliance to a specific Measurement Category defines is that the product won't cause a hazard within those defined synthetic test conditions..... which have been chosen by the committees based on historical data and experience to generally cover the real world situations.
Why is it so hard for people to understand?
The standard is a specific limit that is considered representative of the real world.
There will be real world examples worse than that (but they will be very few/unusual)
There will be real world examples that are benign and never even approach the limit (common, routine, normal, typical, most situations)
Otherwise you end up with an impossible to meet standard that covers a soaking wet gorilla probing a socket outlet outside in the rain, which is connected back to the poles with some obscure 120mm2 cable, during lighting strike on the nearest pole.... and the meter needs to continue operating perfectly afterwards despite being on current mode and connected across the phases.