feat(api): 访谈人格/回复长度策略、口述归一、背景语气与输入净稿全链路

Chat 访谈
- 新增 persona 系统（default / warm_listener / curious_guide）与 background_voice 语气层
- 回复长度由 compute_reply_plan 统一决策（brief / standard / expanded），融合信息密度启发式
- 输入净稿（input_normalize）：编排层可选 rules/llm 归一用户口语后再喂模型与记忆检索
- 记忆证据注入：按用户话检索 memory evidence 并注入 prompt

Memoir 回忆录
- 口述归一（oral_normalize）：segment 原文保留，story 管线取派生净稿作叙事输入
- segment 入队批次门闸：累计字数 + 最长等待秒数，减少零碎提交
- fidelity_check / prompts / narrative_agent 微调
- Alembic 0005：清理跨章节 story 外键

Infra
- Dockerfile 加入 ffmpeg
- pyproject.toml 新增依赖并同步 uv.lock
- .env.example / .env.production 补全新配置项

Tests
- 新增 test_background_voice、test_chat_input_normalize、test_experience_regressions
- 扩展 test_interview_prompts、test_interview_reply_length、test_story_route_oral_invariant

Made-with: Cursor

api/tests/test_chat_input_normalize.py

@@ -0,0 +1,39 @@
+"""聊天输入归一：与 memoir 规则共用，配置独立。"""
+
+from unittest.mock import patch
+
+from app.features.conversation.input_normalize import (
+    apply_conversation_input_rules,
+    normalize_chat_input_for_agent,
+)
+
+
+def test_apply_conversation_rules_matches_memoir_mei_kanshang() -> None:
+    raw = "我去试镜了 美看上我 张伟"
+    assert "没看上我" in apply_conversation_input_rules(raw)
+
+
+def test_normalize_chat_rules_mode() -> None:
+    raw = "美看上我"
+    with patch("app.features.conversation.input_normalize.settings") as m:
+        m.chat_input_normalize_enabled = True
+        m.chat_input_normalize_mode = "rules"
+        m.chat_input_normalize_llm_max_tokens = 512
+        m.chat_input_normalize_llm_max_input_chars = 8000
+        assert normalize_chat_input_for_agent(raw, llm=None) == "没看上我"
+
+
+def test_normalize_chat_disabled_returns_raw() -> None:
+    raw = "美看上我"
+    with patch("app.features.conversation.input_normalize.settings") as m:
+        m.chat_input_normalize_enabled = False
+        m.chat_input_normalize_mode = "rules"
+        assert normalize_chat_input_for_agent(raw, llm=None) == raw
+
+
+def test_normalize_chat_off_mode() -> None:
+    raw = "美看上我"
+    with patch("app.features.conversation.input_normalize.settings") as m:
+        m.chat_input_normalize_enabled = True
+        m.chat_input_normalize_mode = "off"
+        assert normalize_chat_input_for_agent(raw, llm=None) == raw