Add python 3.9 and Gymnasium 1.0 support

[araffin]

See #1
Also reformated code and fix some warnings.

[araffin]

Note for later:
I'm having good results (in less than 100 000 steps) using TQC with Simba:

python train_sbx.py --algo tqc --env frasa-standup-v0 -c hyperparams/simba_tqc.py \
-param n_envs:30 gradient_steps:300 policy_delay:10 --verbose 0 \
--eval-episodes 20 --n-eval-envs 5 -P --vec-env subproc -n 200000

simba_tqc.py (needs araffin/sbx#59)

import optax


default_hyperparams = dict(
    n_envs=1,
    n_timesteps=int(5e5),
    policy="SimbaPolicy",
    # policy="MlpPolicy",
    # learning_rate=3e-4,
    # qf_learning_rate=1e-3,
    policy_kwargs={
        "optimizer_class": optax.adamw,
        "net_arch": {"pi": [128], "qf": [256, 256]},
        # "dropout_rate": 0.01,
        "n_critics": 2,
    },
    learning_starts=10_000,
    normalize={"norm_obs": True, "norm_reward": False},
)

hyperparams = {}

for env_id in [
    "HalfCheetah-v4",
    "Humanoid-v4",
    "HalfCheetahBulletEnv-v0",
    "Ant-v4",
    "Hopper-v4",
    "Walker2d-v4",
    "Swimmer-v4",
    "AntBulletEnv-v0",
    "HopperBulletEnv-v0",
    "Walker2DBulletEnv-v0",
    "BipedalWalkerHardcore-v3",
    "Pendulum-v1",
    "frasa-standup-v0",
]:
    hyperparams[env_id] = default_hyperparams.copy()
    if "Bullet" in env_id:
        hyperparams[env_id].update({"gamma": 0.98})

    if "Swimmer" in env_id:
        hyperparams[env_id].update({"gamma": 0.999})