[LeeLeno]

refactor: rename SAC task g1_sac_wbt → g1_motion_tracking and remove mujoco_deploy.yaml
Task names should be algorithm-agnostic. Also removes the deprecated mujoco_deploy.yaml profile (superseded by g1_wbt_obs).

Summary

• 将 SAC 任务目录 conf/offpolicy/task/sac/g1_sac_wbt/ 重命名为 g1_motion_tracking/，CLI 由 task=sac/g1_sac_wbt/{mujoco,motrix} 改为 task=sac/g1_motion_tracking/{mujoco,motrix}
• 删除 mujoco_deploy.yaml（原通过 --profile deploy 调用的调试训练 profile，已被 g1_wbt_obs 取代），清理 scripts/deploy/export_deploy_config.py、g1_wbt_obs/mujoco.yaml 注释、docs/ 中的 dangling reference
• 同步更新 README.md、docs/users/zh_CN/D-tasks/02-g1-motion-tracking.md、docs/users/zh_CN/E-reference/01-backend-support-matrix.md、tests/benchmark/test_replay_buffer_placement_benchmark.py
• 任务名不再包含算法名，符合"任务名与算法解耦"的设计原则
• YAML 内容字节级未变（仅目录名变化）；task_name: G1MotionTrackingSAC 保持不变，env 类绑定和 checkpoint 目录命名完全一致；Python 代码零改动（从未硬编码过 g1_sac_wbt）

Linked Work

• Issue:
• Milestone:

Validation

• make check
• uv run pytest -m "not slow"
• Hydra 配置解析验证（--cfg job）：sac/g1_motion_tracking/mujoco → task_name: G1MotionTrackingSAC；sac/g1_motion_tracking/motrix → task_name: G1MotionTrackingSAC, sim_backend: motrix；sac/g1_wbt_obs/mujoco → task_name: G1WBTObs
• PPO + MuJoCo 1-iter 烟测：g1_motion_tracking、g1_flip_tracking、g1_wall_flip_tracking 全部通过
• APPO + MuJoCo 1-iter 烟测：g1_motion_tracking 通过（含 checkpoint 保存）
• 全仓库 grep -rn "g1_sac_wbt" 零残留；grep -rn "mujoco_deploy"（不含历史性 mujoco_deploy_v2 注释）零残留
  Commands actually run:

# Hydra config 解析验证
uv run scripts/train_offpolicy.py --cfg job algo=sac task=sac/g1_motion_tracking/mujoco
uv run scripts/train_offpolicy.py --cfg job algo=sac task=sac/g1_motion_tracking/motrix
uv run scripts/train_offpolicy.py --cfg job algo=sac task=sac/g1_wbt_obs/mujoco
# 1-iter 烟测
export HF_ENDPOINT=https://hf-mirror.com/
uv run train --algo ppo --task g1_motion_tracking --sim mujoco algo.num_envs=4 algo.max_iterations=1 training.no_play=true
uv run train --algo ppo --task g1_flip_tracking --sim mujoco algo.num_envs=4 algo.max_iterations=1 training.no_play=true
uv run train --algo ppo --task g1_wall_flip_tracking --sim mujoco algo.num_envs=4 algo.max_iterations=1 training.no_play=true
uv run train --algo appo --task g1_motion_tracking --sim mujoco algo.num_envs=4 algo.max_iterations=1 training.no_play=true
# 残留检查
grep -rn "g1_sac_wbt" .
grep -rn "mujoco_deploy" . | grep -v mujoco_deploy_v2

Impact

• Backend impact: none（YAML 内容未变，仅 config group 路径重命名）
• Platform impact: both
• Training effect expected: no（同一份 YAML，相同 env 类和超参，训练曲线应完全一致）

Artifacts

• W&B:
• benchmark result:
• video / screenshot:
• ONNX / checkpoint:

Checklist

• Added or updated tests where needed（tests/benchmark/test_replay_buffer_placement_benchmark.py 同步更新任务名断言）
• Updated docs if behavior or workflow changed（README + zh_CN docs + backend support matrix）
• Linked the driving issue
• Noted any follow-up work explicitly：使用旧 CLI task=sac/g1_sac_wbt/... 的下游脚本/笔记本需迁移到 task=sac/g1_motion_tracking/...；--profile deploy 已无对应 YAML，需改用 g1_wbt_obs