Narrow down dependency versions

[Max-We]

The dependencies versions in pyproject.toml are not optimized. E.g. gymnasium = ">=0.28.1" which could be lowered. Also, python = ">=3.9,<4.0" could potentially be lowered to 3.7.

To solve this issue, try out the minimal required version for these packages and run the test-suite / Dockerfile and an example training script to validate that the project is still working as expected.