feat: carry the reasoning verdict through to LLMResponse

Both assembly paths fill it, streaming and non-streaming alike. Filling
only one is exactly the divergence this issue exposed: M3 returns
reasoning prose over SSE and nothing at all over the plain endpoint, so
a verdict computed on one path says nothing about the other.

The field defaults to UNKNOWN on both TransportResult and LLMResponse.
A transport that does not judge should not get to declare absence on
the provider's behalf, and a default that stays silent is the only one
that cannot lie.

tests/unit/test_retry.py

@@ -27,6 +27,7 @@ from polygateway.types import (
     GlobalLimits,
     RetryPolicy,
     SourceConfig,
+    ThinkingObservation,
     TransportResult,
 )
 from tests.contracts.conftest import FakeClock
@@ -225,6 +226,29 @@ class TestObservabilityPassthrough:
         assert resp.cached_prompt_tokens is None and resp.model_reported is None
         assert resp.reasoning_tokens is None
 
+    async def test_thinking_observation_reaches_the_response(self):
+        """issue #16/#17: 裁定归 transport,中间件只透传,不得在途中改判。"""
+        result = TransportResult(
+            content="ok",
+            thinking="想一下",
+            prompt_tokens=10,
+            completion_tokens=5,
+            usage_source="measured",
+            ttft_ms=12.0,
+            max_inter_token_ms=3.0,
+            raw={},
+            thinking_observation=ThinkingObservation.OBSERVED,
+        )
+        mw, *_ = _harness([_src("a")], [result])
+        resp = await mw(_REQ)
+        assert resp.thinking_observation is ThinkingObservation.OBSERVED
+
+    async def test_unjudged_transport_result_stays_unknown(self):
+        """不裁定的 transport(如 OCR)透传出来仍是 UNKNOWN,不被默认成 ABSENT。"""
+        mw, *_ = _harness([_src("a")], [_ok()])
+        resp = await mw(_REQ)
+        assert resp.thinking_observation is ThinkingObservation.UNKNOWN
+
 
 class TestRetryAndFailover:
     async def test_transient_switches_source_then_succeeds(self):