I don't see how you think that's what the parent commenter is believing. For starters, the modder didn't make "GTA V 70% faster", he made one small (but noticeable) part 70% better.

And yes, there are many code bases where performance isn't prioritized in all areas, and one totally can expect a skilled engineer coming in to look at performance to find nice speed increases here and there. Not because they are magically better than everyone else, but because a) they are given the freedom to spend time on it where nobody else has been and b) because they are possibly more familiar with approaches and tooling than a dev that hasn't had the opportunity to prioritize performance. All your examples likely do have such people available, and thus bringing someone in is less likely to find something really juicy, unless they have some blindspot. Large game projects for sure have performance experts, but maybe they primarily spend their time on in-game performance and initial resource loading was a blind spot. Maybe that's a common blind spot. And outside games that's commonly waaay worse - performance isn't prioritized, and to be fair the fact that slower-than-necessary software usually sells just as well makes that understandable economically for a long time. But also makes it difficult if it gets to the point where it should/must matter.