Update Lunar lander step to match performance with Gymnasium

[pseudo-rnd-thoughts]

#89 found that mo-lunar-lander had a couple of differences with Gymnasium that caused performance issues.
However, testing the updated version improved performance but still did not match
Therefore, this PR, copy and pasted the Gymnasium version reverting changes that related to the reward vector.
As a result, the mo version now matches step fps (on my laptop)

from gymnasium.utils.performance import benchmark_step
import mo_gymnasium

env = mo_gymnasium.make("LunarLander-v2")
mo_env = mo_gymnasium.make("mo-lunar-lander-v2")

print(f'env benchmark - {benchmark_step(env, target_duration=15)}')  # 16997.796487754913 
print(f'mo-env benchmark - {benchmark_step(mo_env, target_duration=15)}')  # 15513.081151435932