fix: 统一 input_context_expr 计算口径，移除按 api_format 分支的 CASE 逻辑

input_context_expr() 原先按 OpenAI/Gemini 和 Claude 分支计算输入上下文，
现统一为 input_tokens + cache_read_input_tokens，与 usage 表展示口径一致。
同步更新 admin 和 user_me 路由中的缓存命中率注释，并新增单元测试。

tests/unit/test_usage_query.py

@@ -0,0 +1,72 @@
+from __future__ import annotations
+
+from sqlalchemy import create_engine, select
+from sqlalchemy.orm import Session
+
+from src.models.database import Usage
+from src.services.usage.query import input_context_expr
+
+
+def _make_usage(
+    request_id: str,
+    api_format: str,
+    input_tokens: int,
+    cache_read_input_tokens: int,
+) -> Usage:
+    return Usage(
+        request_id=request_id,
+        provider_name="provider-a",
+        model="test-model",
+        api_format=api_format,
+        input_tokens=input_tokens,
+        cache_read_input_tokens=cache_read_input_tokens,
+    )
+
+
+def test_input_context_expr_always_uses_input_plus_cache_read() -> None:
+    engine = create_engine("sqlite:///:memory:")
+    Usage.__table__.create(engine)
+
+    with Session(engine) as session:
+        session.add_all(
+            [
+                _make_usage("openai-ok", "openai:chat", 100, 20),
+                _make_usage("gemini-ok", "gemini:chat", 80, 30),
+                _make_usage("claude-ok", "claude:chat", 100, 20),
+            ]
+        )
+        session.commit()
+
+        rows = session.execute(
+            select(Usage.request_id, input_context_expr().label("ctx")).order_by(Usage.request_id)
+        ).all()
+
+    assert {r[0]: r[1] for r in rows} == {
+        "claude-ok": 120,
+        "gemini-ok": 110,
+        "openai-ok": 120,
+    }
+    engine.dispose()
+
+
+def test_input_context_expr_keeps_same_formula_for_anomalous_openai_compatible_usage() -> None:
+    engine = create_engine("sqlite:///:memory:")
+    Usage.__table__.create(engine)
+
+    with Session(engine) as session:
+        session.add(
+            _make_usage(
+                "openai-anomalous",
+                "openai:chat",
+                input_tokens=10,
+                cache_read_input_tokens=300,
+            )
+        )
+        session.commit()
+
+        total_input_context = session.execute(
+            select(input_context_expr()).where(Usage.request_id == "openai-anomalous")
+        ).scalar_one()
+
+    assert total_input_context == 310
+    engine.dispose()