feat(compare): 比价记录 LLM token 成本落库与展示(按当时价冻结) (#133)

- comparison_record 加 llm_cost_yuan(元/float)+ llm_price_snapshot(JSON)两列
- _backfill_llm_calls 回填时按 app_config 当时单价逐模型算成本、冻结成本+快照到记录
- app_config 新增 llm_token_price 配置(per_model + default 兜底,运营在系统配置页可改)
- services/llm_cost.py:compute_llm_cost 纯函数(按 model 分桶、error/无 usage 跳过、
  脏价格当 unpriced 不抛异常以免连累 token 回填)+ get_llm_prices reader
- admin schema 暴露成本:列表项带 llm_cost_yuan,详情另带价格快照
- tests/test_llm_cost.py(10 测试);scripts/seed_mock_llm_cost.py(mock seeder)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: guke <guke@autohome.com.cn>
Reviewed-on: #133

app/admin/schemas/comparison.py

@@ -36,6 +36,8 @@ class AdminComparisonListItem(BaseModel):
     retry_count: int | None = None
     input_tokens: int | None = None      # Σ usage.prompt_tokens(server 派生)
     output_tokens: int | None = None     # Σ usage.completion_tokens(server 派生)
+    # 本次比价 LLM 总成本(元,按当时价冻结);旧记录/未回填为 None → 前端「成本」列回退估算。见 services/llm_cost.py。
+    llm_cost_yuan: float | None = None
     device_model: str | None = None
     rom_vendor: str | None = None
     rom_name: str | None = None
@@ -72,3 +74,5 @@ class AdminComparisonDetail(AdminComparisonListItem):
     # 原始上报全量;「卡在哪一步」从 raw_payload.platform_results[*].status 读
     # (store_not_found/items_not_found/below_minimum/unsupported = 卡在 找店/加菜/起送/读价)。
     raw_payload: dict | None = None
+    # 算成本所用单价快照 {mode, prices:{model:{...}}}(llm_cost_yuan 继承自列表项)。见 services/llm_cost.py。
+    llm_price_snapshot: dict | None = None