feat: 实现 React 模式循环推理，带超时重试和结构化错误处理

- 更新 intent.py 为 React 模式推理器
- 新增 react_nodes.py: React 模式节点
- 新增 retry_utils.py: 超时和重试工具
- 更新 state.py: 支持循环步数和错误记录
- 重写 subgraph_builder.py: 完整 React 循环流程
- 结构化错误输出，符合 Agent 执行循环最佳实践
- 限制最大推理步数 ≤40，防止无限循环
- RAG 检索带重试和超时保护
- 子图错误可传递给主图处理

backend/app/agent_subgraphs/common/__init__.py

@@ -17,6 +17,7 @@ from .formatter import (
 )
 
 from .intent import (
+    # 旧版 API（保持向后兼容）
     IntentType,
     Intent,
     Entity,
@@ -24,7 +25,17 @@ from .intent import (
     RuleBasedIntentClassifier,
     RuleBasedEntityExtractor,
     IntentRegistry,
-    create_default_intent_parser
+    create_default_intent_parser,
+    # 新版 React 模式 API
+    ReasoningAction,
+    RetrievalConfig,
+    ReasoningResult,
+    BaseIntentReasoner,
+    RuleBasedReactReasoner,
+    LLMReactReasoner,
+    create_react_reasoner,
+    react_reason,
+    get_route_by_reasoning
 )
 
 from .human_review import (
@@ -49,7 +60,7 @@ __all__ = [
     "TemplateManager",
     "OutputRenderer",
     "PresetTemplates",
-    # intent
+    # intent - 旧版
     "IntentType",
     "Intent",
     "Entity",
@@ -58,6 +69,16 @@ __all__ = [
     "RuleBasedEntityExtractor",
     "IntentRegistry",
     "create_default_intent_parser",
+    # intent - 新版 React 模式
+    "ReasoningAction",
+    "RetrievalConfig",
+    "ReasoningResult",
+    "BaseIntentReasoner",
+    "RuleBasedReactReasoner",
+    "LLMReactReasoner",
+    "create_react_reasoner",
+    "react_reason",
+    "get_route_by_reasoning",
     # human_review
     "ReviewStatus",
     "HumanReview",