This is a splendid summary! Much more accessible than the typical review article, even to someone who knows a fair bit about human genetics.

The only thing I'd want to add is that a PRS is usually* a substantial underestimate because it is hard to detect the contribution of many weak alleles. So it's a testament to the polygenicity that the PRS falls well below the heritability, but it doesn't necessarily mean that the polygenic picture isn't the whole story. (I'm also a little bit surprised that you mentioned CNV and not epigenetics--not that either is that robustly verified right now from what I've seen.)

Anyway, height has a similarly large gulf between heritability and variance explained by polygenic score; it's kind of par for the course with the sample sizes we tend to have these days.

(* Technical detail: uncorrected population stratification can lead to large overestimates. So we need to be tentative here, too. Even with height: https://elifesciences.org/articles/39702.)