I don’t consider epidemiology to be bs or just hypothesis generating as do most scientists.

That is a curious statement, science is a procedure that requires falsifiable and testable hypotheses.

So if a lot of high quality epidemiology says saturated fats increase heart disease risk by 20-30% over decades then I will likely believe it.

Ok, so your appealing to consensus? But that is fine for personal beliefs, however, there is evidence supporting saturated fat for long term health.

You should look at a study’s design, strengths, weaknesses etc and evaluate it.

Agreed, a observational study cannot establish cause and effect - by design.

Do you consider astronomy or evolution to be bullshit too because they are based on observation as well?

Astronomy has testable falsifiable hypotheses. As does evolution. You build a model and see if new data you didn't train on fits the model. With plant based nutritional epidemology, people reject new data by leaning things like 'a lot of high quality epidemiology says' and ignoring that contradiction of the model.

These are not impossible things to test. We did it in the 1960s.