Update 8a_train_sqil_sac.ipynb to remove MuJoCo dependency

[AdamGleave]

Bug description

We decided against depending on MuJoCo in CI, but https://github.com/HumanCompatibleAI/imitation/blob/8db6bfb0cfd533f6418e662890499670bd6b010b/docs/tutorials/8a_train_sqil_sac.ipynb has introduced this dependency into our test suite.

Steps to reproduce

Observe tutorial uses HalfCheetah which is a MuJoCo environment.

Suggested resolution: replace with an environment that is native Gymnasium, like [MountainCarContinuous](https://gymnasium.farama.org/environments/classic_control/mountain_car_continuous/] or Pendulum