What I believe is this: people are complex creatures with sophisticated brains. Measuring "ability" objectively is nigh-impossible for a lot of things, but it's not impossible to get a general sense of someone's ability in a particular field.
> What you are dismissing as "flowery language" appears to be a different epistemology from yours. I cannot reconcile "there is a wide range in between" with these concrete categories you are giving.
Think about it like this: there is a color gradient that transitions from blue on the left, to purple in the middle, to red on the right, smoothly. If I point to the leftmost part of the color gradient and ask people what color it is, they would say "blue". Likewise for the right, they would say "red".
In the middle it gets a little more tricky. If start on the left and move in a touch, most people will probably say blue. Move a bit further and some people might say purple, and others blue (would anybody say red? probably not). Move further and now you're in pretty solid purple consensus territory, though you might get the odd person claiming red or blue. You get what I'm trying to say.
There absolutely are children who are "blue" or "red". There aren't a lot, but they are there. Most are purple, and sometimes you can get a sense for whether they're more bluish-purple or reddish-purple. The analogy isn't perfect, because ability is a multidimensional and complex but I think it applies fairly well if you limit your gauge of "red" or "blue" to particular subject matters.
I agree with your assertion that you can't know _everybody's_ category with complete certainty. There are some whose ability or lack of is very apparent, and others that require a bit more inspection.
> You can imagine measurements that provide information about those categories (no one seriously believes a test could directly detect such a thing, however).
A test is one way to help gauge it. But some people are poor test takers. A test doesn't account for someone not sleeping well the night before, or having recently experienced something traumatic. Tests can't effectively differentiate between rote memorization and real understanding, in a lot of cases. You need a good educator to provide a bit more context if you _really_ want to know.
And that's really what it comes down to. Context is everything here. A letter grade can't really tell you if a student is overwhelmed or unmotivated.
I think, when it comes to grades and tests, the outliers I described above will average out over time. A percentage grade alone is not sufficient to provide good direction for a child's education, but a low grade does a pretty decent job of letting you know that something is going wrong.