feat: use SecretStr for inference provider auth credentials

- RemoteInferenceProviderConfig now has auth_credential: SecretStr
- the default alias is api_key (most common name)
- some providers override to use api_token (RunPod, vLLM, Databricks)
- some providers exclude it (Ollama, TGI, Vertex AI)

llama_stack/providers/remote/inference/tgi/config.py

@@ -13,6 +13,8 @@ from llama_stack.schema_utils import json_schema_type
 
 @json_schema_type
 class TGIImplConfig(RemoteInferenceProviderConfig):
+    auth_credential: SecretStr | None = Field(default=None, exclude=True)
+
     url: str = Field(
         description="The URL for the TGI serving endpoint",
     )

llama_stack/providers/remote/inference/tgi/tgi.py

@@ -30,7 +30,7 @@ class _HfAdapter(OpenAIMixin):
     overwrite_completion_id = True  # TGI always returns id=""
 
     def get_api_key(self):
-        return self.api_key.get_secret_value()
+        return "NO KEY REQUIRED"
 
     def get_base_url(self):
         return self.url