I completely agree with everything you said, however I'd like to comment on the last thing you said - Just because the best possible strategy changes depending

I completely agree with everything you said, however I'd like to comment on the last thing you said - Just because the best possible strategy changes depending on the skill of your opponent(s) does not mean that there can't be a strategy that cannot be exploited by your opponent. I think the mistake many people make is that they tend to look at games with a "chess mindset" (there is a single move that is the perfect response to my opponents move) when games with hidden information (such as cs:go) will require a "poker mindset" - a strategy isn't given by a single move but by a probability distribution of actions. In your example, there is an "optimal" way to play this scenario in which you aim at jumping height p% of the time and at normal walking height (1-p)% of the time where the value of p depends on many different factors such as you hitting the headshot if you aim correctly, you hitting it anyways if you don't aim correctly etc. This is not the strategy that will win the the most (if you know for sure that your opponent will jump every time, of course you would never aim at walking height), but it is one that cannot be exploited by your opponent deviating from whatever their optimal strategy is. Of course, improving your skill enough to see these changes in behavior in your opponents takes a long time, which is why we naturally adapt the more exploitative strategies that you mentioned.