fix: integrate PR #3315 review feedback

- envs(robotwin): default `observation_height/width` in
  `create_robotwin_envs` to `DEFAULT_CAMERA_H/W` (240/320) so they
  match the D435 dims baked into `task_config/demo_clean.yml`.
- envs(robotwin): resolve `task_config/demo_clean.yml` via
  `CONFIGS_PATH` instead of a cwd-relative path; works regardless
  of where `lerobot-eval` is invoked.
- envs(robotwin): replace `print()` calls in `create_robotwin_envs`
  with `logger.info(...)` (module-level `logger = logging.getLogger`).
- envs(robotwin): use `_LazyAsyncVectorEnv` for the async path so
  async workers start lazily (matches LIBERO / RoboCasa / VLABench).
- envs(robotwin): cast `agent_pos` space + joint-state output to
  float32 end-to-end (was mixed float64/float32).
- envs(configs): use the existing `_make_vec_env_cls(use_async,
  n_envs)` helper in `RoboTwinEnvConfig.create_envs`; drop the
  `get_env_processors` override so RoboTwin uses the identity
  processor inherited from `EnvConfig`.
- processor: delete `RoboTwinProcessorStep` — the float32 cast now
  happens in the wrapper itself, so the processor is redundant.
- tests: drop the `TestRoboTwinProcessorStep` suite; update the
  mock obs fixture to use float32 `joint_action.vector`.
- ci: hoist `ROBOTWIN_POLICY` and `ROBOTWIN_TASKS` to job-level
  env vars so the task list and policy aren't duplicated across
  eval / extract / parse steps.
- docker: pin RoboTwin + CuRobo upstream clones to commit SHAs
  (`RoboTwin@0aeea2d6`, `curobo@ca941586`) for reproducibility.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/envs/test_robotwin.py

@@ -28,7 +28,6 @@ from unittest.mock import MagicMock, patch
 import gymnasium as gym
 import numpy as np
 import pytest
-import torch
 
 from lerobot.envs.robotwin import (
     ACTION_DIM,
@@ -56,7 +55,7 @@ def _make_mock_task_env(
     """
     obs_dict = {
         "observation": {cam: {"rgb": np.zeros((height, width, 3), dtype=np.uint8)} for cam in cameras},
-        "joint_action": {"vector": np.zeros(ACTION_DIM, dtype=np.float64)},
+        "joint_action": {"vector": np.zeros(ACTION_DIM, dtype=np.float32)},
         "endpose": {},
     }
 
@@ -281,34 +280,3 @@ def test_all_tasks_are_strings():
 
 def test_no_duplicate_tasks():
     assert len(ROBOTWIN_TASKS) == len(set(ROBOTWIN_TASKS))
-
-
-# ---------------------------------------------------------------------------
-# RoboTwinProcessorStep
-# ---------------------------------------------------------------------------
-
-
-class TestRoboTwinProcessorStep:
-    def test_passes_through_images_and_state(self):
-        from lerobot.processor.env_processor import RoboTwinProcessorStep
-        from lerobot.utils.constants import OBS_IMAGES, OBS_STATE
-
-        step = RoboTwinProcessorStep()
-        obs = {
-            f"{OBS_IMAGES}.head_camera": torch.zeros(1, 3, 4, 4),
-            f"{OBS_IMAGES}.left_camera": torch.zeros(1, 3, 4, 4),
-            OBS_STATE: torch.zeros(1, 14),
-        }
-        result = step.observation(obs)
-        assert f"{OBS_IMAGES}.head_camera" in result
-        assert f"{OBS_IMAGES}.left_camera" in result
-        assert result[OBS_STATE].dtype == torch.float32
-
-    def test_state_cast_to_float32(self):
-        from lerobot.processor.env_processor import RoboTwinProcessorStep
-        from lerobot.utils.constants import OBS_STATE
-
-        step = RoboTwinProcessorStep()
-        obs = {OBS_STATE: torch.zeros(1, 14, dtype=torch.float64)}
-        result = step.observation(obs)
-        assert result[OBS_STATE].dtype == torch.float32