feat: 全功能前后端联调完成，47/47 测试通过

前端：
- 新增 DietCapturePage 独立拍照识别页
- 5种消息卡片类型完整实现（数据确认/用药/饮食/报告/快捷选项）
- 任务卡片区：异常警告+数据摘要+自动折叠
- 侧滑抽屉：历史对话列表+对话管理
- 运动计划：进度卡片+创建计划+每日打卡
- 报告页：拍照/相册/PDF上传+分析
- 面板按钮补全血氧/体重录入
- UI 升级：紫色主题+动画+气泡样式
- 全部迁移 Riverpod 3.x API

后端：
- 新增 _UpdateMessageTypeAndMetadata，Tool Calling 自动映射消息类型
- SSE answer 事件携带 type 字段
- 提示词优化（移除"阿福"，语气规则归位）
- 运动计划支持 AI 创建和打卡

测试：
- 新增 full_e2e_test.py 全流程测试（认证/数据CRUD/6个Agent对话/VLM/报告）

backend/src/Health.WebApi/Endpoints/ai_chat_endpoints.cs

@@ -107,6 +107,8 @@ public static class AiChatEndpoints
             var maxIterations = 5;
             var fullResponse = "";
             var completedNormally = false;
+            var messageType = "text";
+            var metadata = new Dictionary<string, object>();
 
             for (int i = 0; i < maxIterations; i++)
             {
@@ -129,7 +131,7 @@ public static class AiChatEndpoints
                             if (!string.IsNullOrEmpty(content))
                             {
                                 fullResponse += content;
-                                await SseWriteAsync(http, new { action = "answer", data = content }, ct);
+                                await SseWriteAsync(http, new { action = "answer", data = content, type = messageType }, ct);
                             }
                         }
                         catch (JsonException) { /* 跳过解析失败的 chunk */ }
@@ -160,6 +162,8 @@ public static class AiChatEndpoints
                         }
                         await SseWriteAsync(http, new { action = "tool_result", tool = tc.Function.Name, data = toolResult }, ct);
 
+                        _UpdateMessageTypeAndMetadata(tc.Function.Name, toolResult, ref messageType, ref metadata);
+
                         messages.Add(new ChatMessage { Role = "tool", Content = JsonSerializer.Serialize(toolResult, JsonOpts), ToolCallId = tc.Id });
                     }
                 }
@@ -597,6 +601,40 @@ public static class AiChatEndpoints
         }
     };
 
+    /// <summary>根据工具调用结果更新消息类型和元数据</summary>
+    private static void _UpdateMessageTypeAndMetadata(string toolName, object toolResult, ref string messageType, ref Dictionary<string, object> metadata)
+    {
+        switch (toolName)
+        {
+            case "record_health_data":
+                messageType = "data_confirm";
+                if (toolResult is IDictionary<string, object> resultDict)
+                {
+                    if (resultDict.TryGetValue("type", out var type))
+                        metadata["type"] = type.ToString();
+                    if (resultDict.TryGetValue("success", out var success) && success is bool b && b)
+                        metadata["success"] = true;
+                }
+                break;
+            case "manage_medication":
+                messageType = "medication_confirm";
+                if (toolResult is IDictionary<string, object> medDict)
+                {
+                    if (medDict.TryGetValue("name", out var name))
+                        metadata["name"] = name.ToString();
+                    if (medDict.TryGetValue("dosage", out var dosage))
+                        metadata["dosage"] = dosage.ToString();
+                }
+                break;
+            case "estimate_food_text":
+                messageType = "diet_analysis";
+                break;
+            case "analyze_report":
+                messageType = "report_analysis";
+                break;
+        }
+    }
+
     /// <summary>压缩图片到合理大小供 VLM API 使用</summary>
     private static void CompressImage(string inputPath, string outputPath, int maxWidth, long quality)
     {