feat(registry): make the Stack query providers for model listing (#2862)

This flips #2823 and #2805 by making the Stack periodically query the
providers for models rather than the providers going behind the back and
calling "register" on to the registry themselves. This also adds support
for model listing for all other providers via `ModelRegistryHelper`.
Once this is done, we do not need to manually list or register models
via `run.yaml` and it will remove both noise and annoyance (setting
`INFERENCE_MODEL` environment variables, for example) from the new user
experience.

In addition, it adds a configuration variable `allowed_models` which can
be used to optionally restrict the set of models exposed from a
provider.

llama_stack/providers/utils/inference/model_registry.py

@@ -20,6 +20,13 @@ from llama_stack.providers.utils.inference import (
 logger = get_logger(name=__name__, category="core")
 
 
+class RemoteInferenceProviderConfig(BaseModel):
+    allowed_models: list[str] | None = Field(
+        default=None,
+        description="List of models that should be registered with the model registry. If None, all models are allowed.",
+    )
+
+
 # TODO: this class is more confusing than useful right now. We need to make it
 # more closer to the Model class.
 class ProviderModelEntry(BaseModel):
@@ -65,7 +72,10 @@ def build_model_entry(provider_model_id: str, model_descriptor: str) -> Provider
 
 
 class ModelRegistryHelper(ModelsProtocolPrivate):
-    def __init__(self, model_entries: list[ProviderModelEntry]):
+    __provider_id__: str
+
+    def __init__(self, model_entries: list[ProviderModelEntry], allowed_models: list[str] | None = None):
+        self.allowed_models = allowed_models
         self.alias_to_provider_id_map = {}
         self.provider_id_to_llama_model_map = {}
         for entry in model_entries:
@@ -79,6 +89,27 @@ class ModelRegistryHelper(ModelsProtocolPrivate):
                 self.alias_to_provider_id_map[entry.llama_model] = entry.provider_model_id
                 self.provider_id_to_llama_model_map[entry.provider_model_id] = entry.llama_model
 
+    async def list_models(self) -> list[Model] | None:
+        models = []
+        for entry in self.model_entries:
+            ids = [entry.provider_model_id] + entry.aliases
+            for id in ids:
+                if self.allowed_models and id not in self.allowed_models:
+                    continue
+                models.append(
+                    Model(
+                        model_id=id,
+                        provider_resource_id=entry.provider_model_id,
+                        model_type=ModelType.llm,
+                        metadata=entry.metadata,
+                        provider_id=self.__provider_id__,
+                    )
+                )
+        return models
+
+    async def should_refresh_models(self) -> bool:
+        return False
+
     def get_provider_model_id(self, identifier: str) -> str | None:
         return self.alias_to_provider_id_map.get(identifier, None)