The quality of evidence usually ranks RCTs and meta analysis of the same as top of the evidence tree. They are evidently very difficult to run because they are expensive and can't be done 'blind' nor be done over the period of time it takes to develop chronic illness.
Food frequency questionnaires are just the tools of the observational study which looks forward or backwards at one cohort of people. Those studies are frequently confounded by the 'Gwyneth Paltrow; effect i.e. healthy people do plenty of healthy things as well as NOT eating meat (non smoking, yoga practice, living in California etc.) - see also The Mormons in Lomalinda CA . The converse is true - that those who truly don't care about dietary advice might be eating red meat in a burger but they're also eating plenty of vegetable oils too and not much salad.
The ancestral health idea does makes sense but isn't as reliable as an RCT because we don't really know what our ancestors ate. We do know they ate a wide range of diets with varying amounts of meat/fish and grains/tubers. The unifying factor seems to be lack of processed foods and very little sugars (fruit in season and honey).
Like those observational studies based on food frequency questionnaires either side can cherry pick their hunter gatherers of choice to make the point for or against red meat. If you like keto you'd go for Kalahari bush men and Eskimos. If you are anti red meat choose Mormons or the Okinawans (though as it turns out they ate quite a lot of pork) or others in the 'Blue Zones'. The only eating pattern not supported by the ancestral health evidence is veganism. I agree with your concerns about the questionnaires being unverifiable!
Interesting question though and I agree that it is worth asking how evidence is gathered in food science!
We do need better reasons than we've been given to stop eating meat especially if you've got metabolic disease.