vllm prompt_logprobs can also be 0

This adjusts the vllm openai_completion endpoint to also pass a
value of 0 for prompt_logprobs, instead of only passing values greater
than zero to the backend.

The existing test_openai_completion_prompt_logprobs was parameterized
to test this case as well.

Signed-off-by: Ben Browning <bbrownin@redhat.com>

llama_stack/providers/remote/inference/vllm/vllm.py

@@ -446,7 +446,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
         model_obj = await self._get_model(model)
 
         extra_body: Dict[str, Any] = {}
-        if prompt_logprobs:
+        if prompt_logprobs is not None and prompt_logprobs >= 0:
             extra_body["prompt_logprobs"] = prompt_logprobs
         if guided_choice:
             extra_body["guided_choice"] = guided_choice