But my initial thoughts are that those are highly cherry picked
YES! All epidemiology is bullshit, by its very nature - the vast majority are cherry picking and polluted by confounders. However, if your standard for compelling evidence is epidemiology you have to account for every study going against your hypothesis.
highest quality epidemiological studies like Framingham
Why do you consider this epidemiology better quality then other epidemiology? It generates a hypothesis that needs to be tested! It has just as many confounders! If you rate quality by alignment with a pre-existing bias, you have to be aware of it.
show eating plants is way better for heart and brain in the long run.
You have to look critically at every study - In what population? In what metabolic context? Enumerate every confounder, every "controlled" variable that has assumptions baked in. Realize FFQs are highly unreliable.