Was there anything like this back then? (Speculative execution, L1 cache, branch prediction all in the same ISA?)

If they were aware, why didn't anyone raise this issue till now?

The people that were aware were probably extracting value out of the fact that not everyone knew.

Could be not worth making a public stink, could be weaponizing the exploit, could be coordinated disclosure.

you mean all these security researchers kept it a secret for the last decade or two?

Until a few days, researchers couldn't prove that this was possible because there was no example of such exploits. But there was the idea that this could be possible. Also this doesn't mean that people with other interests didn't have their own versions of this exploit and kept it a secret.