chore: docs 结构整改 + compose 双目录对齐（合并重建提交）

本提交为 .git 对象库损坏后的重建提交，内容等价于原先三个本地提交
（5e2fd4c2 / 57a53c98 / 5d7e1873）的累积结果，未做任何额外改动。

一、docs 结构整改（整改 #14）
根因：重构时新结构为 untracked 文件，执行 git stash（未带 -u）未纳入，
随后 git reset 拉回 HEAD 旧 tracked 树，导致旧树复活、新旧两棵目录
树并存于 docs/，共 791 文件、双分类体系冲突。

修复动作：
- b2 同名异主题文件改名迁移保全 9 个
- C 类 39 个孤立文件按主题正确归类
- A/B1 类 222 个重复文件删除（新结构已有内容副本）
- 9 个旧独有空目录删除
- 270 处内部引用按 verified 映射改写
- 整改记录 #14 登记于 04-运维文档/部署运维

结果：docs 791 → 569 文件，顶层仅规范 8 类 + 治理文件，单树恢复。
残留：约 20 处指向从未存在文件的陈旧死链，归入独立文档卫生任务。

二、compose 双目录对齐（消除踩坑 A）
- docker-compose.yml：nginx 前端挂载全部由根目录 frontend-*/dist
  改为 src/frontend-*/dist（h5 / agent / admin / terminal）
- docker-compose.dev.yml：dev 服务 build context 与卷同步改 src/
- 效果：本地 docker compose up 不再把根目录 stale dist 挂回，
  与线上一致，分叉隐患消除（已 docker compose config 校验通过）

防复发铁律：
- 重构须提交；仓库修复须 git stash -u 或先 commit
- 新结构须 git add 并提交，避免再次 untracked 复活
- H5 改动只动 src/frontend-h5/，禁改根目录遗留 frontend-*/

docs/02-技术文档/技术架构/sequence-diagram-v1.1.mermaid

@@ -0,0 +1,129 @@
+%% ====================================================================
+%% v1.1 正式技术方案 — 4 个时序图
+%% ====================================================================
+
+%% ========== 时序 ①：点选-广播-落库 ==========
+sequenceDiagram
+    autonumber
+    actor Emp as 员工
+    participant MB as MessageBubble
+    participant Store as ConversationStore
+    participant WS as ws._handle_option_select
+    participant DB as PostgreSQL
+    participant WSM as ws_manager
+    participant Obs as BackendObserver
+    participant Task as process_h5_ai_reply
+
+    Emp->>MB: 点击 option
+    MB->>Store: sendOptionSelect(value, label, sourceId, qId, oId)
+    Note over Store: lastSentOptionContent=label<br/>optionSubmitting=true<br/>5 重 UUID 守卫判定
+    Store->>WS: option_select(6 字段 + client_msg_id)
+    WS->>WS: UUID 校验
+    WS->>DB: BEGIN + pg_advisory_xact_lock
+    WS->>DB: SELECT 5s 内同 UUID
+    alt 5s 内重复命中
+        DB-->>WS: dup row returned
+        WS-->>Store: 无副作用，结束
+    else 首次受理
+        WS->>DB: INSERT Message(msg_type=option_select)
+        DB-->>WS: commit_ts = T1
+        Note over WS,WSM: ★ v1.1 T03: 同步推员工端
+        WS->>WSM: broadcast_to_employees([employee_id], new_message)
+        WSM-->>Store: new_message (employee 端)
+        Store->>Store: trackProcessedMessageId + push message
+        Note over Store: 员工立即看到 ✓ 气泡
+        WS->>WSM: broadcast(new_message) - 坐席
+        WSM-->>Obs: 记录 broadcast_ts = T2
+        Obs-->>Obs: option_select_broadcast_latency_ms = T2 - T1
+        WS->>Task: create_task(process_h5_ai_reply)
+    end
+
+%% ========== 时序 ②：AI 回复回流 ==========
+
+sequenceDiagram
+    autonumber
+    participant Dify as Dify Workflow
+    participant Task as process_h5_ai_reply
+    participant DB as PostgreSQL
+    participant WSM as ws_manager
+    participant WS as H5 useH5WebSocket
+    participant Store as ConversationStore
+    participant MB as MessageBubble
+
+    Dify-->>Task: 返回结构化 reply (ai_text 或 still_thinking)
+    alt Dify 正常返回
+        Task->>DB: INSERT Message(ai_text)
+        DB-->>Task: commit
+    else Dify 超时 ≥ 15s
+        Note over Task: ★ v1.1 T03: still_thinking 兜底<br/>主动补发 option_select_confirm
+        Task->>WSM: broadcast_to_employees(option_select_confirm)
+    end
+    Task->>WSM: broadcast_to_employees(ai_reply)
+    WSM->>WS: ai_reply { type: "ai_reply", data: {...} }
+    WS->>Store: handleAiReply(data)
+    Note over Store: ★ v1.1 Bug 4 软校验:<br/>conversation_id 不匹配 → console.warn<br/>但不再 return<br/>仅 message_id 已去重才 return
+    Store->>Store: trackProcessedMessageId(message_id)
+    Store->>Store: messages.value.push(finalMessage)
+    Store-->>MB: reactive 触发重渲
+    MB->>MB: 显示 AI 答案气泡（打字机逐字）
+    Note over MB: ★ v1.1 T02 派生:<br/>isOptionSelected 从 messages 过滤<br/>不再依赖内存 ref
+
+%% ========== 时序 ③：视觉分组 ==========
+
+sequenceDiagram
+    autonumber
+    actor Emp as 员工
+    participant Store as ConversationStore
+    participant Panel as ChatPanel
+    participant MB as MessageBubble
+
+    Note over Store: messages 序列示例:<br/>[ai_structured Q1]<br/>[option_select Q1.optA] 第 1 次<br/>[ai_text 回复1]<br/>[option_select Q1.optB] 第 2 次（重选）<br/>[ai_text 回复2]
+
+    Emp->>Store: 选 Q1.optB
+    Store->>Store: sendOptionSelect → WS send
+    Note over Store: groupedMessagesByQuestion (v1.1 ★)<br/>按 question_id 相邻聚合<br/>折叠 Q1.optA，激活 Q1.optB
+
+    Panel->>MB: v-for msg in groupedMessages
+    MB->>MB: option_select 模板加 data-question-id
+    alt collapsed (历史同题选项)
+        MB->>MB: class=message-bubble--option-select collapsed<br/>背景灰底 + 折叠图标 ▾
+    else active (本次选项)
+        MB->>MB: class=message-bubble--option-select active<br/>蓝底 + ✓
+    end
+    MB->>MB: ai_text 紧随其后，按时间顺序独立出现
+    Note over MB: 3 个 UI 元素顺序独立:<br/>① ✓当前 ② AI 思考占位 ③ AI 答案
+
+%% ========== 时序 ④：时延监控 ==========
+
+sequenceDiagram
+    autonumber
+    actor Emp as 员工 (Chromium A)
+    actor Agent as 坐席 (Chromium B)
+    participant H5Emp as H5 useH5WebSocket
+    participant H5Agt as H5 useH5WebSocket
+    participant WS as ws._handle_option_select
+    participant WSM as ws_manager
+    participant Obs as BackendObserver
+    participant Script as measure-option-latency.mjs
+
+    Script->>Script: 启动双 context + login 双方
+    Script->>Script: t0_start = performance.now()
+
+    Emp->>H5Emp: click option
+    H5Emp->>WS: option_select payload (含 client_msg_id UUID)
+    H5Emp->>H5Emp: 记录 client_send_ts = T0
+
+    WS->>WSM: broadcast_to_employees + broadcast
+    WSM->>H5Agt: new_message(option_select)
+    H5Agt->>H5Agt: 渲染 ✓ 气泡
+    H5Agt->>Script: postMessage { event: 'rendered', ts: T_agent }
+
+    H5Emp->>H5Emp: 收到自己的 new_message (员工端)
+    H5Emp->>Obs: POST /api/backend-observer/record<br/>{metric: 'option_select_e2e_latency_ms',<br/>value: T_agent - T0}
+    WSM->>Obs: 记录 option_select_persist_latency_ms (commit_ts)<br/>记录 option_select_broadcast_latency_ms (broadcast_ts)
+
+    Script->>Script: t1_end = performance.now() (员工点击 → 坐席渲染)
+    Script->>Obs: GET /api/backend-observer/metrics?name=option_select_*
+    Obs-->>Script: 4 个指标样本
+    Script->>Script: p50/p95 计算 + 告警 (>100ms warning)
+    Script->>Script: 输出 report.json