this problem is quite common. This is neither a measurement error nor a machine error. The computer does everything right.
The problem lies more with the engineer who thinks that the large plane has a secondary meaning.
in reality, the cylinder A is much too short to represent the main axis (specify the vector)
imagine you are trying to check the problem with a hand measuring device (e.g. a steel angle). I doubt you see a slit of light within the .25 height.
(datum A is only to centering in 99% of these cases)
you can get around the problem by evaluating the concentricity to B|A instead of just to A
I've already experienced that the "designer" can be very stubborn and doesn't see it, but it might be useful if you ask him if you're allowed to do it that way.