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/inline/inference/sentence_transformers/sentence_transformers.py

@@ -20,6 +20,7 @@ from llama_stack.apis.inference import (
     ToolDefinition,
     ToolPromptFormat,
 )
+from llama_stack.apis.models import ModelType
 from llama_stack.providers.datatypes import Model, ModelsProtocolPrivate
 from llama_stack.providers.utils.inference.embedding_mixin import (
     SentenceTransformerEmbeddingMixin,
@@ -41,6 +42,8 @@ class SentenceTransformersInferenceImpl(
     InferenceProvider,
     ModelsProtocolPrivate,
 ):
+    __provider_id__: str
+
     def __init__(self, config: SentenceTransformersInferenceConfig) -> None:
         self.config = config
 
@@ -50,6 +53,22 @@ class SentenceTransformersInferenceImpl(
     async def shutdown(self) -> None:
         pass
 
+    async def should_refresh_models(self) -> bool:
+        return False
+
+    async def list_models(self) -> list[Model] | None:
+        return [
+            Model(
+                identifier="all-MiniLM-L6-v2",
+                provider_resource_id="all-MiniLM-L6-v2",
+                provider_id=self.__provider_id__,
+                metadata={
+                    "embedding_dimension": 384,
+                },
+                model_type=ModelType.embedding,
+            ),
+        ]
+
     async def register_model(self, model: Model) -> Model:
         return model