Change HILSerlRobotEnvConfig to inherit from EnvConfig

Added support for hil_serl classifier to be trained with train.py
run classifier training by python lerobot/scripts/train.py --policy.type=hilserl_classifier
fixes in find_joint_limits, control_robot, end_effector_control_utils

lerobot/common/datasets/utils.py

@@ -412,7 +412,7 @@ def dataset_to_policy_features(features: dict[str, dict]) -> dict[str, PolicyFea
 
             names = ft["names"]
             # Backward compatibility for "channel" which is an error introduced in LeRobotDataset v2.0 for ported datasets.
-            if names[2] in ["channel", "channels"]:  # (h, w, c) -> (c, h, w)
+            if names is not None and names[2] in ["channel", "channels"]:  # (h, w, c) -> (c, h, w)
                 shape = (shape[2], shape[0], shape[1])
         elif key == "observation.environment_state":
             type = FeatureType.ENV