Automatic setup does not cover GPU --cuda mode as the piper1-gpl setup does not require the onnxruntime-gpu dependency.

The failure only occurs at runtime and falls back to CPU which is not obvious until you inspect the logs and see
"/.venv/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py:121: UserWarning: Specified provider 'CUDAExecutionProvider' is not in available provider names.Available providers: 'AzureExecutionProvider, CPUExecutionProvider'"