test(streaming): keep pytest.raises to one statement, drop the spend read for an empty stream

An empty stream writes no spend row on main either, so the [DONE]-only
cell checks only what the client sees.

tests/integration/streaming/test_stream_clean_eof_without_finish_reason.py

@@ -202,11 +202,11 @@ def test_chat_stream_that_sends_only_done_still_completes_empty(gateway: Gateway
     """Gemini's OpenAI-compatible endpoint answers with only [DONE] when reasoning uses up max_tokens."""
     with wire_server(_upstream("openai")) as wire, gateway.scenario() as scenario:
         model: Final = scenario.model(model=_DEPLOYMENTS["openai"], api_base=wire.url, api_key=_PROVIDER_KEY)
-        text, call_id = _chat(gateway, model, "ending-done-only")
+        text, _ = _chat(gateway, model, "ending-done-only")
         assert _chat_text(text) == "", text
         assert _finish_reasons(text) == ("stop",), text
         assert _chat_errors(text) == (), text
-        assert spend_row(call_id)["status"] == "success"
+        assert text.rstrip().endswith("data: [DONE]"), text
 
 
 @pytest.mark.parametrize("family", ["openai", "anthropic", "gemini", "langgraph"])

tests/unit/litellm_core_utils/test_streaming_handler.py

@@ -1577,7 +1577,6 @@ async def test_mid_stream_error_keeps_credentials_in_the_api_base_out(routed_api
     assert "deployment-secret" not in debug_info
 
 
-
 def _incomplete_openai_chat_chunk() -> ModelResponseStream:
     return ModelResponseStream(
         id="chatcmpl-incomplete-eof",
@@ -7686,10 +7685,7 @@ async def test_no_synthetic_finish_reason_logged_when_provider_sent_none(sync_mo
         logging_obj=logging_obj,
     )
     with pytest.raises(MidStreamFallbackError):
-        if sync_mode:
-            list(response)
-        else:
-            [c async for c in response]
+        await _collect(response, sync_mode)
 
     assert response.received_finish_reason is None
     assert all(not (c.choices and c.choices[0].finish_reason) for c in response.chunks)