Roger, as one who works in colourimetrics in a hybridised workflow, I draw your attention to the following statement:Isn't there an inherent conflict between saturation and color accuracy at some point, where to make the color more saturated it has to be more purely one wavelength, whereas the original was more mixed?
"Additionally, many of the "colors" within Lab space fall outside the gamut of human vision, and are therefore purely imaginary; these "colors" cannot be reproduced in the physical world."
The garish rendering of the image on the left is an oversimplification of the way LAB tweaks the gamut (saturation and brightness) to try and approximate human vision, but we do not see the view this way. The colour space is just awful for practical use and certainly should be avoided with films like Velvia, Provia, what remains of the Kodak emulsions and anything with high RGB values. I've noticed that LAB is routinely applied to photographs in Royal Auto, (e.g. AAA in other countries). The colourisation is terrible, a joke.
Kodachrome was not a very saturated film to my eyes. The many hundreds of slides I have in archives could be described as bright and lively, but not saturated. If saturation was required, a polariser would be slapped on, an in many landscape images I have of the 80s and early 90s, that's what I did.
That Wiki article, while informative and well researched, is bogged down in a welter of grand technical methodology and resource that in modern colourimetric work is totally unnecessary beyond basics because so much of the work is automated (it has to be; photographers and artists would never get any work done puddling around all this!). An example is the deep CIELAB-CIEXYZ conversions and the incredibly foggy CIELAB explanation.