I don't see how this is practical. You could spend a lifetime testing the work of others and still not get through it all, let alone get to working on anything original. Progress is made by building on the work of others.
This seems like a complete utter waste of time.
In real life most life impacting academic research is much more right than wrong. You are far better served assuming so. Unless you want to waste your time going back to basic science and rebuilding all the academic knowledge in most things you wish to do.
So it’s more like suppose you want to paint your house green, and you read that somebody says you can mix red and blue paint to make a really cool green paint. Instead of immediately going out and buying enough red and blue paint to cover your whole house, first buy a small amount of red and blue paint, mix them together, and see if you get that neat green paint.
It’s common sense, but the window dressings of academia can lead you to burn time and money on things that are totally silly because somebody important-sounding said they did it once.
Where people get burned is that there’s an enormous power imbalance—-junior scientists can end up stuck trying and failing to make green paint out of red and blue paint because nobody senior is going to take them seriously if they can’t make green paint. This presents a serious ethical challenge if making green paint is impossible.
It's fair to question things, especially if they don't make sense to you and even if acknowledged authorities are behind them. However, (1) something that you may question is not necessarily something I may question, and (2) questioning may be a waste of time.
If a paper that says mixing red and blue paint makes green paint has a thousand citations, perhaps you don't need to question it because others already have. If you can't reproduce it, the simplest thing to do is ask an expert who says it is possible to do it.
If making green paint is impossible I think that it will eventually self correct, or is simply inconsequential. In some instances it may take a while, but if the alternative is to reprove a result before using it — that seems like something only a fool would do or someone with infinite time.