fix: relax knowledge retrieval usage schema

api/services/entities/knowledge_retrieval_inner.py

@@ -173,14 +173,6 @@ class InnerKnowledgeRetrieveRequest(BaseModel):
 class InnerKnowledgeRetrieveUsage(ResponseModel):
     """Serialized LLM usage payload returned by dataset retrieval."""
 
-    model_config = ConfigDict(
-        from_attributes=True,
-        extra="forbid",
-        populate_by_name=True,
-        serialize_by_alias=True,
-        protected_namespaces=(),
-    )
-
     prompt_tokens: int
     completion_tokens: int
     total_tokens: int

api/tests/unit_tests/services/test_knowledge_retrieval_inner_service.py

@@ -98,6 +98,8 @@ class TestInnerKnowledgeRetrievalService:
             "total_price": "0",
             "currency": "USD",
             "latency": 0,
+            "time_to_first_token": None,
+            "time_to_generate": None,
         }
         mock_rag_cls.return_value = rag