`.[agents]` extra omits `langchain-anthropic`, so a shipped example cannot run

**Files:** `pyproject.toml` (`[tool.poetry.extras]`), affects `examples/agents/langgraph/43_react_agent_multi_model.py`

**Symptom:** After the documented install — `pip install -e '.[agents]'`, which is what CI runs — the example fails immediately:
```
ModuleNotFoundError: No module named 'langchain_anthropic'
```

**Cause:** The `agents` extra lists `langchain`, `langchain-core`, `langchain-openai`, `langgraph`, `google-adk`, `openai`, `openai-agents`, `anthropic`, `claude-code-sdk` — but not `langchain-anthropic`, which `43_react_agent_multi_model.py` imports for `ChatAnthropic`.

**Fix:** Add `langchain-anthropic` to the `agents` extra (and its own `langchain-anthropic` extra alongside the existing `langchain` / `langgraph` ones).

**Verify:** From a clean env, `pip install -e '.[agents]'` then `python examples/agents/langgraph/43_react_agent_multi_model.py` runs. Confirmed: with `langchain-anthropic 1.5.5` installed manually it passes on both server types (5.8s / 76.7s).

Update the [tool.poetry.extras] entries in pyproject.toml, adding langchain-anthropic to the agents extra and its own extra. Start by checking the dependency declarations, then run pip install -e '.[agents]' in a clean environment and execute examples/agents/langgraph/43_react_agent_multi_model.py; done means the example runs without ModuleNotFoundError.

由索引模型根据 Issue 内容生成。