Current research suggests that while salt is bad for people WITH hypertension the causal link between elevated consumption and hypertension isn't what was earlier believed. One study here https://www.medicalnewstoday.com/articles/317099.php
I guess the key is to live a balanced life, don't consume anything in excess and don't worry too much. "In medium virtus est"
Another piece of guidance I like is this (paraphrased): Eat absolutely anything you want. But make it all yourself. If you want cookies and cake and ice cream, great! Buy flour and sugar and butter and chocolate and milk and cream, and make them yourself.[1]
Now, these don't directly address your point about conflicting research. They give reasonable heuristics that work well, though. Now, clearly you can come up with pathological examples that follow the advice above - any of us could come up with dozens that are reasonable. But being precise and 100% correct is not the point of a heuristic.
[0] https://michaelpollan.com/reviews/how-to-eat/
[1] Paraphrased from an interview in Cooked. The 4-part series was produced for Netflix.
The aphorism "science advances one funeral at a time" is saddeningly true.
The more rigorous areas are those with little time lag between input and output, and results that are easily verifiable. Do you die within 2 weeks of taking this drug or are you cured - strong results. Take this pill every day for 40 years to reduce your all factor mortality by 15% - not so much. An imploded sphere of plutonium either gives you a lump of metal or an earth shattering kaboom - no need for a second lab to reproduce.
Areas with high time lag and difficult verification/reproduction are more vulnerable to research and policy entrepreneurialism. Whether it's salt hypertension, red meat and heart disease, the Cornell Brand Lab, or much of sociology errors last for decades because of the high costs of verification and the low rewards.
You'll notice that nearly EVERYTHING that you consume comes with the State of California Cancer warning.