You know, there's a lot of really good data on Stack Overflow, regardless of all the issues.
I hope somebody saves it all.
Up until a few years ago, SO published full dumps onto the Internet Archive. There has been a community effort which has picked up the torch and is continuing to do the same on a ~monthly cadence.
For better or for worse, the contents of Stack Overflow are literally the reason LLM coding has taken over the world.
They used to publish a regular torrent of everything, not sure if they still do that? Maybe you could grab yourself a copy for posterity.
Perhaps not what you intended, but most of it is stored in various models' weights.