The author presents a four-color primary set. It clearly has its roots in Helmholtz's color physiology, but I confess I don't quite understand how it would be used. For me, I would miss the important blue/orange pairing.
So the author is correct that there are four opponent colors, but not correct that they can all be arranged on a wheel so easily.
Color spaces are three dimensional geometries, attempts to reduce them to two dimensions are always wrong just like every 2D map projection creates distortions.
Yellow is not lighter than blue, though I can understand why you say this.
In the hue, saturation and lightness (hsl) understanding of color, hue terms such as Blue/yellow and lightness terms such as light/dark are separate entities.
It is perfectly possible to have a blue and yellow of the same lightness value, but the yellow would be more dull. More dull = less pure = less saturated. The peak saturation to lightness of different hues is not consistent. Yellow is most yellow (ie most saturate/pure) when it is quite light. Conversly the peak saturationn of green or blue is in the dark range.
As you rightfully state, One key issue with the color wheel is that it is two dimensional. There does not exist a remotely complete color model that does not recognize at least three dimensions.
I am not dissing the utility of the color wheel, but it has application only to the ordering of hues.