chore: support default model in moderations API (#3890)

# What does this PR do?
https://platform.openai.com/docs/api-reference/moderations supports
optional model parameter.

This PR adds support for using moderations API with model=None if a
default shield id is provided via safety config.

## Test Plan
added tests

manual test:
```
> SAFETY_MODEL='together/meta-llama/Llama-Guard-4-12B'   uv run llama stack run starter
> curl http://localhost:8321/v1/moderations \
  -H "Content-Type: application/json" \
  -d '{
    "input": [
        "hello"
    ]
  }'
```

llama_stack/distributions/template.py

@@ -24,6 +24,7 @@ from llama_stack.core.datatypes import (
     DistributionSpec,
     ModelInput,
     Provider,
+    SafetyConfig,
     ShieldInput,
     TelemetryConfig,
     ToolGroupInput,
@@ -188,6 +189,7 @@ class RunConfigSettings(BaseModel):
     default_datasets: list[DatasetInput] | None = None
     default_benchmarks: list[BenchmarkInput] | None = None
     vector_stores_config: VectorStoresConfig | None = None
+    safety_config: SafetyConfig | None = None
     telemetry: TelemetryConfig = Field(default_factory=lambda: TelemetryConfig(enabled=True))
     storage_backends: dict[str, Any] | None = None
     storage_stores: dict[str, Any] | None = None
@@ -290,6 +292,9 @@ class RunConfigSettings(BaseModel):
         if self.vector_stores_config:
             config["vector_stores"] = self.vector_stores_config.model_dump(exclude_none=True)
 
+        if self.safety_config:
+            config["safety"] = self.safety_config.model_dump(exclude_none=True)
+
         return config