[Rebase] Rebase to vllm v0.19.0 (#2475)

tests/entrypoints/openai_api/test_serving_speech.py

@@ -1664,6 +1664,7 @@ class TestFishSpeechServing:
         assert all(allowed_special is None for _, _, allowed_special in tokenizer.calls)
 
     def test_build_fish_clone_prompt_normalizes_text_fields(self, fish_speech_server):
+        fish_speech_server._fish_speech_tokenizer = _FakeFishTokenizer()
         fish_speech_server._estimate_fish_prompt_len = MagicMock(return_value=123)
 
         request = OpenAICreateSpeechRequest(

tests/entrypoints/test_async_omni_abort.py

@@ -13,8 +13,8 @@ def test_generate_accepts_request_after_repeated_cancellations():
         submitted_request_ids = []
         aborted_request_batches = []
 
-        async def fake_add_request_async(*, request_id, prompt, sampling_params_list, final_stage_id):
-            del prompt, sampling_params_list, final_stage_id
+        async def fake_add_request_async(*, request_id, prompt, sampling_params_list, final_stage_id, **kwargs):
+            del prompt, sampling_params_list, final_stage_id, kwargs
             submitted_request_ids.append(request_id)
 
         async def fake_abort_async(request_ids):

tests/entrypoints/test_omni_entrypoints.py

@@ -113,6 +113,7 @@ class FakeAsyncOmniEngine:
         sampling_params_list: list[Any] | None = None,
         final_stage_id: int = 0,
         arrival_time: float | None = None,
+        **kwargs: Any,
     ) -> None:
         msg = {
             "request_id": request_id,