Can you point to anywhere in the 5E books where it states that?
They are not going to waste valuable book space to tell you that 'nothing changed'! When you turn on the TV news, the newscaster tells you about new stuff. For example, it might be in the news that France has got itself a new president, but he doesn't waste half the broadcast time by telling us all the counties who have not changed presidents since yesterday!
The whole rulebook is like that. Each element it describes (race/class/feat/spell) tells us what this element alters. It only tells us what the element doesn't alter if the lack of such clarity would cause ambiguity.
In order for D&D 5E to remove the assumption of the 3d6 bell curve it would have to tell us what the 5E assumption is! If it did that then any old assumption would fall away.
Even so, there is no way that 5E would simply say something like, "NPCs are
not rolled up on 3d6 in order, if that is what you're thinking!" They would simply print the new rule of what it now is.
They didn't. On top of this, every piece of evidence that does get printed in 5E regarding ability scores for average NPCs remains consistent with that 3d6 bell curve assumption.
If we compare the two sets of evidence, we might think that there is no
certainty for that assumption, given the lack of specific verbiage. But crucially, there is far more evidence for that assumption than for any competing assumption.
I don't deny that logically ability scores should be based on some kind of bell curve, but if we look at IQ, a little more than 50% of the population is "average". If we have no good way of modeling a bell curve, it becomes pretty irrelevant.
Gygax wrote a mini essay concerning that very thing in the 1E DMG. In it, he mentioned that some players might think of the D&D Int stat as being IQ/10, or as a creature's IQ as being its Int score x 10. He was keen to point out that this is
not the case; that IQ and Int measure related but different things (Int covers more things). Upshot being that the bell curve which applies to real world IQ scores simply is not the same bell curve that applies to D&D Int, or any other D&D ability score.
If there is a perceived disparity between the bell curves of D&D Int and real world IQ, it's not that the D&D bell curve is 'wrong', it's that it is wrong to assume that D&D Int is the same thing as real world IQ.