The paleo movement has long since moved away from "eat what paleoliths ate" to focus on "nothing in biology makes sense except in light of evolution".[1] Multiple thought leaders (e.g. Chriss Kresser[2] and Robb Wolf) have acknowledged genetic changes such as lactase persistence and changes in iron metabolism that have arisen in the last 10,000 years. Succesful inclusion of dietary grains however, requires fundamental changes in our digestive tract. From Paul Jaminet's "The Perfect Health Diet" pg 196:
"Grasses evolved in concert with grazing mammals: both originated at the same and they became common together. [...] To reproduce successfully despite being regularly eaten, grasses evolved [...] toxic compounds specifically designed to sabotage digestion in mammals. [...] Grazing mammals have evolved defenses for these toxins-for instance, digestive organs such as rumens that allow the brunt of the toxins to be taken by bacteria." [3]
So things like gluten, opioids, agglutinin, and adiponectin trigger autoimmune reactions, neuropathy, accelerate degenerative diseases, and prevent absorption of micronutrients.
I find it ironic OP admits the studies of mummified remains were agricultural societies and then asserts "certainly atherosclerosis was common even among hunter-gatherers." Analysis of modern hunter gatherers and the most reliable accounts of hunter gatherer populations (researchers that actually lived among those they studied) attest to remarkably low occurrences of diseases of civilization while eating diets devoid of grains and high in saturated fat. [4] Rather than a "fascination with the noble savage", paleo eating is about optimizing food quality (i.e. low toxins) and approximating the macronutrient ratios of hunter-gatherer populations. Certainly a segment of the population may thrive on a grain based diet of 70-80% carbohydrate, but that isn't the case for many.
As far as 'science based medicine' is concerned regarding paleo, there is one meaningful study I'm aware of. [5] However, I highly encourage a look at Robb Wolf's recent post,[1] a fantastic summary of the strengths EMB has (antiviral, trauma) compared to the weaknesses (degenerative diseases) and nutrition. He writes that most of what happens in medical practice is based off observation anyway, not double blind placebo controlled RCTs. I'm not against "evidence based" but it's not the only thing out there. Manipulating statistics [6] [7] and noncontextual studies are easily misleading. After experimenting (including a brief bout of vegetarianism) I've been eating paleo for about two years and am the leanest and most energetic I've ever been. The best way to solve the nutrition challenge is to change variables until you are happy with your health.
[1] http://robbwolf.com/2013/03/15/evidence-based-medicine-fraud...
My favorite line from this post is "based of this wacky stuff called the best data we had at the time"
[2] http://chriskresser.com/rhr-what-science-really-says-about-t...
[3] http://perfecthealthdiet.com/notes/#Ch19
[4] http://perfecthealthdiet.com/notes/#Ch2
[5] http://www.staffanlindeberg.com/DiabetesStudy.html
[6]http://www.zoeharcombe.com/2013/03/meat-consumption-and-mort...
[7] "The Great Cholesterol Con" by Malcolm Kendrick and "Good Calories Bad Calories" by Gary Taubes are thorough analyses of the problems in the field of nutrition.