Support Jax environments

[adzcai]

By now there's a growing community of researchers using Jax for reinforcement learning. It's a growing community and there's a vast range of environment APIs:

• https://github.com/RobertTLange/gymnax
• https://github.com/dunnolab/xland-minigrid
• https://github.com/google/brax
• https://github.com/sotetsuk/pgx
• https://github.com/instadeepai/jumanji
• https://github.com/EdanToledo/Stoa

etc. As it seems unlikely that most libraries will change their APIs in the future, it would be fantastic to have a central location for converting between them. Given Farama Foundation's central role in the RL community, this repo seems like a convenient place to support converting between different APIs, instead of having each project hosting different wrappers, possibly with differing functionality (e.g. auto reset semantics). Would this be a possibility? I'd be happy to contribute.

[pseudo-rnd-thoughts]

Hey @adzcai, I have a different plan to provide a Gymnasium style API that is jax-first which if adopted would allow everyone to use the same implementation and be co-used like Gymnasium is for numpy based RL

[adzcai]

That sounds super exciting! I'd love to hear more -- is there a work in progress I could contribute to?