feat: track logical call statistics across governed calls

src/polygateway/client.py

@@ -9,6 +9,7 @@
 from __future__ import annotations
 
 import asyncio
+import dataclasses
 import hashlib
 import json
 import random
@@ -46,6 +47,7 @@ from polygateway.types import (
     Effort,
     LLMResponse,
     TelemetryStatus,
+    _CallContext,
     coerce_effort,
     validate_caller_dimensions,
     validate_request_overlay,
@@ -289,6 +291,10 @@ class GatewayClient:
         self._structured_available = structured_strategy is not None
         self._terminal = terminal  # 内部引用: 装配自省/测试用
         self._handler = compose(middlewares, terminal)
+        # 逻辑调用统计需要同一只注入钟(1.3.5);现之前只传给中间件未自存
+        self._now = now
+        # 终态行由公开边界统一写出(T3),故边界也需持有 emitter
+        self._emitter = emitter
         self._transport = transport
         self._telemetry = telemetry
         self._cache = cache
@@ -370,6 +376,8 @@ class GatewayClient:
             else coerce_effort(reasoning_effort, origin="chat(reasoning_effort=...)")
         )
         validate_thinking_raw(sampling, effort=effort, wire=None, origin="chat overlay")
+        # 三项校验均已通过 → 进入统计边界(设计 §3: 输入校验异常在边界之外,保持原行为)
+        context = _CallContext(now=self._now)
         request = ChatRequest(
             messages=messages,
             session_id=session_id,
@@ -383,8 +391,11 @@ class GatewayClient:
             reasoning_effort=effort,
             tenant_id=dimension_tenant_id,
             meta=dimensions,
+            call_context=context,
         )
-        return await self._handler(request)
+        response = await self._handler(request)
+        # 快照在返回前冻结: 故它含缓存命中路径与已完成的内联遥测耗时
+        return dataclasses.replace(response, call_stats=context.snapshot())
 
     async def aclose(self) -> None:
         """幂等释放**自建**资源: transport、遥测、缓存、限流/熔断后端。