Fix experimental normalize reward wrapper

[raphajaner]

Description

Should resolve #271 in a minimal way with simple tests.

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• I have run the pre-commit checks with pre-commit run --all-files (see CONTRIBUTING.md instructions to set it up)
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[pseudo-rnd-thoughts]

We are changing the VectorEnv to not inherit from Env due to their differences in step data. Therefore, we will have different wrappers for Env and VectorEnv so could you remove the vector code

[raphajaner]

Okay I see, then it's actually enough to change just a single line:

self.discounted_reward: np.array = np.array([0.0])

which has been a float before.