feat(chat): 实现人脸智能聊天核心功能

- 新增小程序人脸聊天控制器 AppChatController，支持会话创建、消息收发、历史查询及会话关闭
- 集成智谱 GLM 模型客户端 GlmClient，支持流式文本生成与回调
- 新增聊天会话与消息实体类及 MyBatis 映射，实现数据持久化
- 提供 FaceChatService 接口及实现，封装聊天业务逻辑包括同步/流式消息发送
- 引入 zai-sdk 依赖以支持调用智谱 AI 大模型能力
- 支持基于人脸 ID 的唯一会话管理与用户权限校验
- 消息记录包含角色、内容、追踪 ID 及延迟信息，便于调试与分析

src/main/java/com/ycwl/basic/controller/mobile/AppChatController.java

@@ -0,0 +1,98 @@
+package com.ycwl.basic.controller.mobile;
+
+import com.ycwl.basic.model.jwt.JwtInfo;
+import com.ycwl.basic.model.mobile.chat.*;
+import com.ycwl.basic.service.mobile.FaceChatService;
+import com.ycwl.basic.utils.ApiResponse;
+import com.ycwl.basic.utils.JwtTokenUtil;
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.http.MediaType;
+import org.springframework.web.bind.annotation.*;
+import org.springframework.web.servlet.mvc.method.annotation.ResponseBodyEmitter;
+
+import java.util.concurrent.CompletableFuture;
+
+/**
+ * 小程序人脸智能聊天接口。
+ */
+@Slf4j
+@RestController
+@RequestMapping("/api/mobile/chat/v1")
+@RequiredArgsConstructor
+public class AppChatController {
+
+    private final FaceChatService faceChatService;
+
+    /**
+     * 获取或创建会话（同一人脸只保留一条）。
+     */
+    @PostMapping("/faces/{faceId}/conversation")
+    public ApiResponse<ChatConversationVO> createConversation(@PathVariable Long faceId) {
+        JwtInfo worker = JwtTokenUtil.getWorker();
+        ChatConversationVO vo = faceChatService.getOrCreateConversation(faceId, worker.getUserId());
+        return ApiResponse.success(vo);
+    }
+
+    /**
+     * 同步发送消息，适用于短回复或前端自行轮询。
+     */
+    @PostMapping("/conversations/{conversationId}/messages")
+    public ApiResponse<ChatSendMessageResp> sendMessage(@PathVariable Long conversationId,
+                                                        @RequestBody ChatSendMessageReq req) {
+        JwtInfo worker = JwtTokenUtil.getWorker();
+        ChatSendMessageResp resp = faceChatService.sendMessage(conversationId, worker.getUserId(),
+                req.getContent(), req.getTraceId());
+        return ApiResponse.success(resp);
+    }
+
+    /**
+     * 流式返回，使用 HTTP chunked。小程序侧用 wx.request 的 onChunkReceived 消费。
+     */
+    @PostMapping(value = "/conversations/{conversationId}/messages/stream", produces = "text/plain;charset=UTF-8")
+    public ResponseBodyEmitter streamMessage(@PathVariable Long conversationId,
+                                             @RequestBody ChatSendMessageReq req) {
+        JwtInfo worker = JwtTokenUtil.getWorker();
+        ResponseBodyEmitter emitter = new ResponseBodyEmitter(30_000L);
+        CompletableFuture.runAsync(() -> {
+            try {
+                faceChatService.sendMessageStream(
+                        conversationId,
+                        worker.getUserId(),
+                        req.getContent(),
+                        req.getTraceId(),
+                        chunk -> {
+                            try {
+                                emitter.send(chunk, new MediaType("text", "plain", java.nio.charset.StandardCharsets.UTF_8));
+                            } catch (Exception e) {
+                                emitter.completeWithError(e);
+                            }
+                        });
+                emitter.complete();
+            } catch (Exception e) {
+                log.error("streamMessage error", e);
+                emitter.completeWithError(e);
+            }
+        });
+        return emitter;
+    }
+
+    /**
+     * 查询历史消息，cursor 为最后一条 seq，limit 为条数。
+     */
+    @GetMapping("/conversations/{conversationId}/messages")
+    public ApiResponse<ChatMessagePageResp> listMessages(@PathVariable Long conversationId,
+                                                         @RequestParam(value = "cursor", required = false) Integer cursor,
+                                                         @RequestParam(value = "limit", required = false) Integer limit) {
+        JwtInfo worker = JwtTokenUtil.getWorker();
+        ChatMessagePageResp resp = faceChatService.listMessages(conversationId, cursor, limit, worker.getUserId());
+        return ApiResponse.success(resp);
+    }
+
+    @PostMapping("/conversations/{conversationId}/close")
+    public ApiResponse<String> closeConversation(@PathVariable Long conversationId) {
+        JwtInfo worker = JwtTokenUtil.getWorker();
+        faceChatService.closeConversation(conversationId, worker.getUserId());
+        return ApiResponse.success("OK");
+    }
+}