refactor(backend): 启动环境初始化与资源项目化（零跨项目依赖）

- main.py：uv run main.py 一键启动——读取 backend/.env、TORCH_HOME/NLTK_DATA 指向项目内缓存、certifi SSL + NO_PROXY 网络直连、默认开启 s2s，启动横幅
- vendor/：补丁后 s2s 云化栈移入项目（backend/vendor/s2s-cloud），pyproject sources 改相对路径
- .gitignore：忽略 .torch-cache / nltk_data 缓存
- AGENTS.md：进程管理铁律（AI 不启动/重启服务，由用户操作）

backend/vendor/s2s-cloud/AGENTS.md

@@ -0,0 +1,26 @@
+# Repository Instructions
+
+- Never include `codex` in branch names or pull request titles.
+- Keep release pull requests focused on version metadata and release documentation.
+- Do not commit local build artifacts such as `dist/`, `build/`, or generated wheel/sdist files.
+
+## Publishing to PyPI
+
+PyPI publishing is handled by GitHub Actions in `.github/workflows/publish.yml`. The workflow runs on pushed tags that match `v*`, builds the package with `uv build`, checks the artifacts with `twine check --strict`, and publishes through the configured `pypi` environment.
+
+To prepare a release:
+
+1. Confirm the intended version is not already published on PyPI.
+2. Bump `version` in `pyproject.toml`.
+3. Bump `__version__` in `src/speech_to_speech/__init__.py`.
+4. Open and merge a pull request with only the release preparation changes.
+
+To publish after the release PR is merged:
+
+1. Update `main` locally: `git checkout main && git pull origin main`.
+2. Create an annotated tag for the version: `git tag -a vX.Y.Z -m "Release vX.Y.Z"`.
+3. Push the tag: `git push origin vX.Y.Z`.
+4. Watch the `Publish` GitHub Actions workflow complete successfully.
+5. Verify the new version appears at `https://pypi.org/project/speech-to-speech/`.
+
+Only upload manually if the GitHub Actions workflow is unavailable and the maintainers have explicitly chosen that fallback.