Gymnasium v1.2.0

In Gym v0.24, v4 MuJoCo environments were added that used a different simulator (mujoco, not mujoco-py).
Having been 3 years since v0.24 and with mujoco-py being unmaintained, it is limiting our ability to support Python 3.13.
Therefore, in this release, we have moved the MuJoCo v2 and v3 to the Gymnasium-Robotics project, meaning the users who cannot upgrade to the v4 or v5 MuJoCo environments should update their code to:

import gymnasium as gym
import gymnasium_robotics  # `pip install "gymnasium-robotics[mujoco-py]"`

gym.register_envs(gymnasium_robotics)  # optional

env = gym.make("Humanoid-v3")

In addition, we have added support for Python 3.13 (and dropped Python 3.8 and 3.9 following NumPy and other projects).

New Features

• Add AddWhiteNoise and ObstructView wrappers that add noise to RGB renderings either across the whole image or sections by @sparisi (#1243)
• Adds rainy and fickle options for Taxi environment that were implemented in the original paper by @foreverska (#1315)
• Add wrappers.ArrayConversion, a generic conversion wrapper between Array API compatible frameworks (like NumPy, Torch, Jax, etc) by @amacati (#1333)
• Add garbage collector on RecordVideo wrapper to prevent out of memory problems by @vicbentu (#1378)

Bug Fixes

• Fix MuJoCo add_markers for mujoco>=3.2 by @amacati (#1329)
• Update CliffWalking version in envs registry by @abouelsaadat (#1334)
• Fix jax 0.6.0 regression in jax_to_numpy by @GaetanLepage (#1361)
• Fix Atari Preprocessing to only check env has NOOP action when noop_max>0 by @pseudo-rnd-thoughts (#1393)
• Fix all warnings in tests/spaces by @amacati (#1396)
• Don't override SDL's audio driver on Linux by @AnErrupTion (#1387)

In addition, this release includes numerous updates to the documentation, most importantly to the introductory pages, with an aim to make them easier for new users of Gymnasium or RL to understand.

Full Changelog: v1.1.1...v1.2.0