For example for Monero, I read that the guy who optimized the hash algorithm, first mined $150,000 of XMR for himself
To be fair that was a bit of a special case, as the code was deliberately deoptimized and obfuscated, then taken over by a group that wasn't familiar with it at all (and had many different aspects of the code to be concerned with, not just PoW), and then became popular and skyrocketed in value quickly.
Removing any of these elements would have made the windfall much smaller. Certainly removing the deliberate deoptimization and obfuscation should be easy.
Anyway, I'm not sure that changes any of your conclusions, but generalizing from Monero is likely wrong as it was a special case.