feat: use SecretStr for inference provider auth credentials (#3724)

# What does this PR do?

use SecretStr for OpenAIMixin providers

- 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)

addresses #3517 

## Test Plan

ci w/ new tests

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

@@ -21,11 +21,6 @@ class LlamaProviderDataValidator(BaseModel):
 
 @json_schema_type
 class LlamaCompatConfig(RemoteInferenceProviderConfig):
-    api_key: str | None = Field(
-        default=None,
-        description="The Llama API key",
-    )
-
     openai_compat_api_base: str = Field(
         default="https://api.llama.com/compat/v1/",
         description="The URL for the Llama API server",

llama_stack/providers/remote/inference/llama_openai_compat/llama.py

@@ -21,9 +21,6 @@ class LlamaCompatInferenceAdapter(OpenAIMixin):
     Llama API Inference Adapter for Llama Stack.
     """
 
-    def get_api_key(self) -> str:
-        return self.config.api_key or ""
-
     def get_base_url(self) -> str:
         """
         Get the base URL for OpenAI mixin.