fix(question_gen): gates reject_reason returns raw reason; short-circuit skips LLM

- GateReport.reject_reason now returns gate.reason directly (no [name] prefix)
- verbatim short-circuit sets other 3 gates to SKIP without calling LLM
- test_high_verbatim_shortcircuits asserts zero LLM calls and SKIP verdicts

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

app/question_gen/gates.py

@@ -131,14 +131,9 @@ class GateReport:
     @property
     def reject_reason(self) -> str | None:
         """首个 FAIL 门的 reason，全部通过返回 None。"""
-        for name, gate in [
-            ("key_verify", self.key_verify),
-            ("blind_answer", self.blind_answer),
-            ("multi_true", self.multi_true),
-            ("leak_test", self.leak_test),
-        ]:
+        for gate in (self.key_verify, self.blind_answer, self.multi_true, self.leak_test):
             if gate.verdict == GateVerdict.FAIL:
-                return f"[{name}] {gate.reason}"
+                return gate.reason
         return None
 
 
@@ -419,29 +414,27 @@ async def run_gates(
     返回:
         GateReport 四门汇总。
     """
-    # Phase 1: verbatim 前置短路
+    # Phase 1: verbatim 前置短路 — 不调 LLM，直接返回
     if postprocess.verbatim_ratio > _VERBATIM_THRESHOLD:
         logger.info(
-            "verbatim_ratio={:.3f} > {:.1f}，key_verify 短路 FAIL",
+            "verbatim_ratio={:.3f} > {:.1f}，key_verify 短路 FAIL，其余三门 SKIP",
             postprocess.verbatim_ratio,
             _VERBATIM_THRESHOLD,
         )
-        key_verify_result = GateResult(
-            verdict=GateVerdict.FAIL,
-            reason=f"verbatim_ratio={postprocess.verbatim_ratio:.3f} exceeds threshold {_VERBATIM_THRESHOLD}",
+        skip_result = GateResult(
+            verdict=GateVerdict.SKIP,
+            reason="skipped due to verbatim short-circuit",
             raw_response="",
         )
-        # 其余三门仍然并发执行（收集完整诊断信息）
-        blind_result, multi_result, leak_result = await asyncio.gather(
-            _gate_blind_answer(candidate, llm, session_id=session_id),
-            _gate_multi_true(candidate, tree, llm, session_id=session_id),
-            _gate_leak_test(candidate, family_spec, llm, session_id=session_id),
-        )
         return GateReport(
-            key_verify=key_verify_result,
-            blind_answer=blind_result,
-            multi_true=multi_result,
-            leak_test=leak_result,
+            key_verify=GateResult(
+                verdict=GateVerdict.FAIL,
+                reason=f"verbatim_ratio={postprocess.verbatim_ratio:.3f} exceeds threshold {_VERBATIM_THRESHOLD}",
+                raw_response="",
+            ),
+            blind_answer=skip_result,
+            multi_true=skip_result,
+            leak_test=skip_result,
         )
 
     # Phase 2: 四门并发执行

tests/unit/test_gates.py

@@ -293,7 +293,7 @@ class TestRunGates:
 
     @pytest.mark.asyncio()
     async def test_high_verbatim_shortcircuits(self) -> None:
-        """verbatim_ratio > 0.5 → key_verify 直接 FAIL，不调用 LLM。"""
+        """verbatim_ratio > 0.5 → key_verify 直接 FAIL，其余 SKIP，不调用 LLM。"""
         llm = MockLLM([_make_llm_response("pass", "should not be called")])
         candidate = _make_candidate()
         tree = _make_tree()
@@ -310,3 +310,9 @@ class TestRunGates:
         assert report.passed is False
         assert report.key_verify.verdict == GateVerdict.FAIL
         assert "verbatim" in report.key_verify.reason.lower()
+        # 其余三门应为 SKIP
+        assert report.blind_answer.verdict == GateVerdict.SKIP
+        assert report.multi_true.verdict == GateVerdict.SKIP
+        assert report.leak_test.verdict == GateVerdict.SKIP
+        # LLM 不应被调用
+        assert len(llm.calls) == 0