refactor: 统一 API 格式显示函数，补全 Usage/Trace 的 provider 链路信息

- 提取 formatApiFormat 为共享工具函数，替换各组件中分散的 API_FORMAT_LABELS 调用
- Trace API 返回密钥认证类型(key_auth_type)和 OAuth 套餐信息(key_oauth_plan_type)
- 请求时间线展示密钥认证方式标签（OAuth/Vertex AI/Kiro 等）
- Usage 记录补充 provider_id/endpoint_id/key_id，避免 curl 复现时缺失 provider 信息
- curl 复现兜底从 RequestCandidate 表查找 provider 信息
- Headers Diff 面板左右独立滚动并同步垂直滚动位置

src/services/usage/_recording_helpers.py

@@ -256,10 +256,13 @@ def update_existing_usage(
     existing_usage.actual_total_cost_usd = usage_params["actual_total_cost_usd"]
     existing_usage.rate_multiplier = usage_params["rate_multiplier"]
 
-    # 更新 Provider 侧追踪信息
-    existing_usage.provider_id = usage_params["provider_id"]
-    existing_usage.provider_endpoint_id = usage_params["provider_endpoint_id"]
-    existing_usage.provider_api_key_id = usage_params["provider_api_key_id"]
+    # 更新 Provider 侧追踪信息（仅在有新值时更新，避免覆盖已有数据）
+    if usage_params.get("provider_id"):
+        existing_usage.provider_id = usage_params["provider_id"]
+    if usage_params.get("provider_endpoint_id"):
+        existing_usage.provider_endpoint_id = usage_params["provider_endpoint_id"]
+    if usage_params.get("provider_api_key_id"):
+        existing_usage.provider_api_key_id = usage_params["provider_api_key_id"]
 
     # 更新元数据（如 billing_snapshot/dimensions 等）
     if usage_params.get("request_metadata") is not None:

src/services/usage/telemetry.py

@@ -172,6 +172,10 @@ class MessageTelemetry:
         response_body: dict[str, Any] | None = None,
         response_headers: dict[str, Any] | None = None,
         client_response_headers: dict[str, Any] | None = None,
+        # Provider 侧追踪信息（用于 curl 复现等场景）
+        provider_id: str | None = None,
+        provider_endpoint_id: str | None = None,
+        provider_api_key_id: str | None = None,
         # 格式转换追踪
         endpoint_api_format: str | None = None,
         has_format_conversion: bool = False,
@@ -183,9 +187,6 @@ class MessageTelemetry:
         """
         记录失败请求
 
-        注意：Provider 链路信息（provider_id, endpoint_id, key_id）不在此处记录，
-        因为 RequestCandidate 表已经记录了完整的请求链路追踪信息。
-
         Args:
             input_tokens: 预估输入 tokens（来自 message_start，用于中断请求的成本估算）
             output_tokens: 预估输出 tokens（来自已收到的内容）
@@ -228,6 +229,10 @@ class MessageTelemetry:
             client_response_headers=client_response_headers,
             response_body=response_body or {"error": error_message},
             request_id=self.request_id,
+            # Provider 侧追踪信息
+            provider_id=provider_id,
+            provider_endpoint_id=provider_endpoint_id,
+            provider_api_key_id=provider_api_key_id,
             # 模型映射信息
             target_model=target_model,
             # 请求元数据
@@ -254,6 +259,10 @@ class MessageTelemetry:
         response_body: dict[str, Any] | None = None,
         response_headers: dict[str, Any] | None = None,
         client_response_headers: dict[str, Any] | None = None,
+        # Provider 侧追踪信息
+        provider_id: str | None = None,
+        provider_endpoint_id: str | None = None,
+        provider_api_key_id: str | None = None,
         # 格式转换追踪
         endpoint_api_format: str | None = None,
         has_format_conversion: bool = False,
@@ -294,6 +303,10 @@ class MessageTelemetry:
             client_response_headers=client_response_headers,
             response_body=response_body or {},
             request_id=self.request_id,
+            # Provider 侧追踪信息
+            provider_id=provider_id,
+            provider_endpoint_id=provider_endpoint_id,
+            provider_api_key_id=provider_api_key_id,
             target_model=target_model,
             metadata=request_metadata,
         )