fix: remove --voxcpm2-base-lm/acoustic args (not valid for llama-server CLI)

llama-server does not have voxcpm2 CLI args. TTS model paths are passed
via omni_init JSON body (model_dir/tts_bin_dir).

modal_deploy/deploy_omni.py

@@ -171,8 +171,6 @@ def serve():
         llama_server_bin,
         "-m", paths["main"],
         "--mmproj", paths["vision"],
-        "--voxcpm2-base-lm", paths["tts_base_lm"],
-        "--voxcpm2-acoustic", paths["tts_acoustic"],
         "--host", "127.0.0.1",
         "--port", str(LLAMA_SERVER_PORT),
         "-ngl", "99",
@@ -894,8 +892,6 @@ def test_omni():
         ls_bin,
         "-m", paths["main"],
         "--mmproj", paths["vision"],
-        "--voxcpm2-base-lm", paths["tts_base_lm"],
-        "--voxcpm2-acoustic", paths["tts_acoustic"],
         "--host", "127.0.0.1",
         "--port", str(LLAMA_SERVER_PORT),
         "-ngl", "99",

tmp_omni_output.txt

@@ -0,0 +1,25 @@
+✓ Initialized. View run at 
+https://modal.com/apps/andrew-jiabin/main/ap-Oe0oWTtAGqGqJ6ZS8Pesj5
+✓ Created objects.
+├── 🔨 Created mount PythonPackage:modal_deploy
+├── 🔨 Created mount 
+│   C:\Users\Andre\codes\LJB\hackthon\for_qclaw\PregoPal\modal_deploy\llamacpp_
+│   omni
+├── 🔨 Created web function serve => 
+│   https://andrew-jiabin--prego-pal-minicpm-omni-serve-dev.modal.run
+├── 🔨 Created function diagnose_volume.
+├── 🔨 Created function test_inference.
+└── 🔨 Created function test_omni.
+[PregoPal] ========== TEST OMNI (llama-server built-in endpoints) ==========
+[PregoPal] main: /models/MiniCPM-o-4_5-gguf/MiniCPM-o-4_5-Q4_K_M.gguf (exists=True)
+[PregoPal] vision: /models/MiniCPM-o-4_5-gguf/vision/MiniCPM-o-4_5-vision-F16.gguf (exists=True)
+[PregoPal] audio: /models/MiniCPM-o-4_5-gguf/audio/MiniCPM-o-4_5-audio-F16.gguf (exists=True)
+[PregoPal] tts_base_lm: /models/MiniCPM-o-4_5-gguf/tts/MiniCPM-o-4_5-tts-F16.gguf (exists=True)
+[PregoPal] tts_acoustic: /models/MiniCPM-o-4_5-gguf/tts/MiniCPM-o-4_5-projector-F16.gguf (exists=True)
+[PregoPal] token2wav_dir: /models/MiniCPM-o-4_5-gguf/token2wav-gguf (exists=True)
+  Starting llama-server with omni support...
+[FAIL] llama-server not ready after 90s
+[llama-server stderr] (1 lines):
+  error: invalid argument: --voxcpm2-base-lm
+✓ App completed. View run at 
+https://modal.com/apps/andrew-jiabin/main/ap-Oe0oWTtAGqGqJ6ZS8Pesj5