feat(api+app): 对话阶段化、回忆录流水线与客户端会话体验

- DB: segments 用户输入文本（Alembic 0002）
- Chat: 阶段检测/阶段提示/回复限制，编排与访谈/画像 prompts 调整
- Memoir: 忠实度检查 agent，叙事与分类等链路更新
- Core: agent 日志、Alembic 启动、LangChain/日志/配置等
- Story: time_hints；Memory 检索与相关测试
- Expo: 助手头像、会话页与消息拆分、实时会话与文案/i18n
- Docs/scripts/tests: 迁移脚本、LLM JSON/记忆检索文档、新增单测

api/tests/test_classification_fragment.py

@@ -5,6 +5,7 @@ import pytest
 from app.agents.memoir.classification_agent import (
     ClassificationAgent,
     _looks_like_fragment_only,
+    _parse_category_from_llm_response,
 )
 
 
@@ -32,6 +33,19 @@ def test_classify_skips_story_for_birth_year_without_llm() -> None:
     assert agent.classify("1999年出生", fallback_stage="childhood", llm=None) is None
 
 
+@pytest.mark.parametrize(
+    "raw,expected",
+    [
+        ('{"category": "childhood"}', "childhood"),
+        ('```json\n{"category": "none"}\n```', "none"),
+        ("childhood", "childhood"),
+        ('"education"', "education"),
+    ],
+)
+def test_parse_category_from_llm_response(raw: str, expected: str) -> None:
+    assert _parse_category_from_llm_response(raw) == expected
+
+
 def test_classify_fallback_when_no_llm_and_narrative_snippet() -> None:
     agent = ClassificationAgent()
     out = agent.classify(