[Edge][WP5] Keep DELTA outputs for talker-only (codec-token) deployments

The speech path forced FINAL_ONLY for Qwen3-TTS deployments without async
chunks, so a talker-only server emitted one engine output per request and all
codec frames arrived at once (measured: 1 engine output per request, first
codes at output #1). A talker-only stage (final_output + latent) now keeps
DELTA outputs so codec frames stream per step.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

tests/entrypoints/openai_api/test_codec_stream_protocol.py

@@ -181,3 +181,21 @@ def test_input_cancel_mid_stream_aborts_request(mocker):
                 assert payload["cancelled"] is True
         assert "codec.done" in types and types[-1] == "session.done"
     svc.engine_client.abort.assert_awaited()
+
+
+@pytest.mark.core_model
+@pytest.mark.cpu
+def test_talker_only_deployment_detection():
+    from types import SimpleNamespace
+
+    from vllm_omni.entrypoints.openai.serving_speech import OmniOpenAIServingSpeech
+
+    probe = OmniOpenAIServingSpeech.__new__(OmniOpenAIServingSpeech)
+    probe._tts_stage = SimpleNamespace(final_output=True, final_output_type="latent")
+    assert probe._is_talker_only_deployment() is True
+    probe._tts_stage = SimpleNamespace(final_output=False, final_output_type="latent")
+    assert probe._is_talker_only_deployment() is False
+    probe._tts_stage = SimpleNamespace(final_output=True, final_output_type="audio")
+    assert probe._is_talker_only_deployment() is False
+    probe._tts_stage = None
+    assert probe._is_talker_only_deployment() is False