feat: let one call ask for a different tier than its source defaults to

The three-layer priority (call > source > enable_thinking sugar > silence)
now lives in one pure function, thinking.effective_effort(). The assembly
guard and the request hot path used to each carry their own inline copy of
the sugar conversion; two copies of the same judgement drift into the worst
shape there is - passes at assembly, raises at runtime.

The guard now also honours effort_fallback, so a source that opted into
nearest is no longer sentenced at assembly for a tier it could have mapped.

tests/unit/test_thinking.py

@@ -13,6 +13,7 @@ from polygateway.thinking import (
     DEFAULT_CAPABILITIES,
     ThinkingCapability,
     ThinkingUnsupportedError,
+    effective_effort,
     get_capability,
     observe_thinking,
     reconcile_thinking,
@@ -616,3 +617,67 @@ class TestCapabilityTierList:
         assert get_capability("MiniMax-M2.7").can_disable is False
         assert get_capability("MiniMax-M2.5").can_disable is False
         assert get_capability("MiniMax-M3").can_disable is True
+
+
+class TestEffectiveEffort:
+    """三层优先级的**唯一**判定处(设计 §4.2): 请求级 > 源级 > 语法糖 > 不表态。
+
+    收口成一个纯函数,是因为它此前在装配守卫与 transport 里各写了一份就地转换:
+    两份各自演化的判定,迟早会在"装配期放行、运行期报错"这种最难查的形态上分叉。
+    """
+
+    def test_request_beats_source(self):
+        assert (
+            effective_effort(
+                request_effort=Effort.MAX, source_effort=Effort.LOW, enable_thinking=None
+            )
+            is Effort.MAX
+        )
+
+    def test_source_beats_sugar(self):
+        assert (
+            effective_effort(request_effort=None, source_effort=Effort.HIGH, enable_thinking=None)
+            is Effort.HIGH
+        )
+
+    def test_none_request_does_not_clear_source(self):
+        """请求级"没表态"绝不能被读成"要求关闭"——那会静默改掉源级的默认档。"""
+        assert (
+            effective_effort(request_effort=None, source_effort=Effort.LOW, enable_thinking=None)
+            is Effort.LOW
+        )
+
+    def test_request_none_tier_is_an_opinion(self):
+        """`Effort.NONE` 是一次明确的表态,必须压过源级档位而不是被当成缺省。"""
+        assert (
+            effective_effort(
+                request_effort=Effort.NONE, source_effort=Effort.MAX, enable_thinking=None
+            )
+            is Effort.NONE
+        )
+
+    def test_enable_thinking_true_is_auto(self):
+        """`True` → `auto`(开启但不指定强度),而**不是**旧版硬编码的 medium。"""
+        assert (
+            effective_effort(request_effort=None, source_effort=None, enable_thinking=True)
+            is Effort.AUTO
+        )
+
+    def test_enable_thinking_false_is_the_none_tier(self):
+        assert (
+            effective_effort(request_effort=None, source_effort=None, enable_thinking=False)
+            is Effort.NONE
+        )
+
+    def test_sugar_is_the_last_word_only(self):
+        """语法糖排在最末: 显式配了档位就以档位为准(矛盾组合已被构造期挡下)。"""
+        assert (
+            effective_effort(request_effort=None, source_effort=Effort.LOW, enable_thinking=True)
+            is Effort.LOW
+        )
+
+    def test_all_absent_is_no_opinion(self):
+        """三层都不表态 → None(随模型默认),与 `Effort.NONE` 严格区分。"""
+        assert (
+            effective_effort(request_effort=None, source_effort=None, enable_thinking=None) is None
+        )