Bob, I'm not quite sure what you are asking.
First off, grit size has nothing to do with surface flatness.
Second, stated flatness tolerances are exactly that, a given hone, surface plate etc. will be within the quoted total deviance range of x thousands or x micron from perfectly flat which would be 0 deviance. That means it could be less, but never more than what is stated, and most samples will be less (based on the proven Bell curve) than the quoted maximum, provided quality control is up to snuff of course.
All this flatness business amounts to little more than bragging rights in so far as honing razors go. An 8" long hone that is off by 0.001" from perfect flatness will not perform any worse than a perfect one for razor sharpening due to the miniscule size of the razors bevel. As the size of whatever you are honing becomes larger, such a variation *can* begin to affect the outcome. Shapton offers their high quality reference plates to ensure a very quick progression through the grit sizes when honing those large bevels as seen on Japanese plane blades and the like... those bevels can be 1/2" deep and 2" + wide... if there were a 0.001" hollow in that bevel from honing on a 1K hone that was slightly out, it would take you forever to remove that hollow using the next grit size, normally a 4K hone. As you would progress up in grit size, it would just get worse as those uber fine hones don't remove much material at all... they just polish the surface.
There is also the issue of the user. Using these hones and reference plates or flattening stones by hand will often introduce more unevenness from incorrect use than the tolerances of the equipment would indicate.
In the end, you want to have all your hones that you would normally use in progression to be of similar shape... if all your hones were dished in the center by 3 thousands of an inch, it would not normally be a problem as each hone is the same. Where you run into issues is when one hone is dished, the next one is crowned, the third one is skewed etc. and even then it simply means you might have to add some more time on the later hones as they have to remove any minor deviations from the previous hone.
I hope this is clearer than it seems to me right now.... my head hurts.
Extreme accuracy has it's place in high end optics, measuring devices, laser optics, astronomical telescopes etc. but those are very specialized endeavors where minute errors are magnified many thousands of times until they become glaring faults.
Regards
Christian