Make Gym environment have customizable action and state spaces

[duyminh1998]

I should not have to create a new environment each time I want to train a different agent. We need to look into implementing custom wrappers or a custom constructor that would allow us to create Gymnasium environments for different simulation scenarios with different state and action spaces.