refactor(stream): 将不完整流 token 估算逻辑收敛到 StreamContext

- 新增 has_partial_response / ensure_estimated_output_tokens / should_estimate_incomplete_tokens 方法
- CancelledError 路径在归因前即补充 output_tokens，确保日志包含估算值
- CLI Handler 和 Chat Handler 的兜底估算统一使用 should_estimate_incomplete_tokens
- 移除 cli_monitor_mixin 和 stream_telemetry 中重复的条件判断
- 新增对应单元测试

tests/api/handlers/base/test_stream_telemetry.py

@@ -0,0 +1,66 @@
+from __future__ import annotations
+
+import time
+from types import SimpleNamespace
+from unittest.mock import AsyncMock
+
+import pytest
+
+from src.api.handlers.base import stream_telemetry as stream_telemetry_module
+from src.api.handlers.base.stream_context import StreamContext
+from src.api.handlers.base.stream_telemetry import StreamTelemetryRecorder
+
+
+class _DummyDb:
+    def close(self) -> None:
+        pass
+
+
+@pytest.mark.asyncio
+async def test_record_stream_stats_estimates_tokens_for_failed_partial_stream(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    recorder = StreamTelemetryRecorder(
+        request_id="req-telemetry",
+        user_id="1",
+        api_key_id="2",
+        client_ip="127.0.0.1",
+        format_id="openai:chat",
+    )
+    recorder._get_telemetry_writer = AsyncMock(  # type: ignore[method-assign]
+        return_value=SimpleNamespace(include_bodies=False)
+    )
+    recorder._dispatch_record = AsyncMock()  # type: ignore[method-assign]
+    recorder._update_candidate_status = AsyncMock()  # type: ignore[method-assign]
+
+    ctx = StreamContext(
+        model="test-model",
+        api_format="openai:chat",
+        request_id="req-telemetry",
+        user_id=1,
+        api_key_id=2,
+    )
+    ctx.provider_name = "test-provider"
+    ctx.status_code = 503
+    ctx.data_count = 2
+    ctx.chunk_count = 4
+    ctx.append_text("partial output")
+
+    monkeypatch.setattr(stream_telemetry_module, "get_db", lambda: iter([_DummyDb()]))
+    monkeypatch.setattr(
+        stream_telemetry_module.SystemConfigService,
+        "should_log_body",
+        lambda _db: False,
+    )
+    monkeypatch.setattr(stream_telemetry_module.config, "stream_stats_delay", 0)
+
+    await recorder.record_stream_stats(
+        ctx,
+        original_headers={},
+        original_request_body={"input": [{"content": "hello world"}]},
+        start_time=time.time(),
+    )
+
+    assert ctx.input_tokens > 0
+    assert ctx.output_tokens == max(1, len("partial output") // 4)
+    recorder._dispatch_record.assert_awaited_once()  # type: ignore[attr-defined]