chore(cleanup)!: remove tool_runtime.rag_tool (#3871)

Kill the `builtin::rag` tool group completely since it is no longer
targeted. We use the Responses implementation for knowledge_search which
uses the `openai_vector_stores` pathway.

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

llama_stack/providers/registry/inference.py

@@ -42,6 +42,7 @@ def available_providers() -> list[ProviderSpec]:
             # CrossEncoder depends on torchao.quantization
             pip_packages=[
                 "torch torchvision torchao>=0.12.0 --extra-index-url https://download.pytorch.org/whl/cpu",
+                "numpy tqdm transformers",
                 "sentence-transformers --no-deps",
                 # required by some SentenceTransformers architectures for tensor rearrange/merge ops
                 "einops",

llama_stack/providers/registry/tool_runtime.py

@@ -7,33 +7,13 @@
 
 from llama_stack.providers.datatypes import (
     Api,
-    InlineProviderSpec,
     ProviderSpec,
     RemoteProviderSpec,
 )
-from llama_stack.providers.registry.vector_io import DEFAULT_VECTOR_IO_DEPS
 
 
 def available_providers() -> list[ProviderSpec]:
     return [
-        InlineProviderSpec(
-            api=Api.tool_runtime,
-            provider_type="inline::rag-runtime",
-            pip_packages=DEFAULT_VECTOR_IO_DEPS
-            + [
-                "tqdm",
-                "numpy",
-                "scikit-learn",
-                "scipy",
-                "nltk",
-                "sentencepiece",
-                "transformers",
-            ],
-            module="llama_stack.providers.inline.tool_runtime.rag",
-            config_class="llama_stack.providers.inline.tool_runtime.rag.config.RagToolRuntimeConfig",
-            api_dependencies=[Api.vector_io, Api.inference, Api.files],
-            description="RAG (Retrieval-Augmented Generation) tool runtime for document ingestion, chunking, and semantic search.",
-        ),
         RemoteProviderSpec(
             api=Api.tool_runtime,
             adapter_type="brave-search",

llama_stack/providers/registry/vector_io.py

@@ -119,7 +119,7 @@ Datasets that can fit in memory, frequent reads | Faiss | Optimized for speed, i
 #### Empirical Example
 
 Consider the histogram below in which 10,000 randomly generated strings were inserted
-in batches of 100 into both Faiss and sqlite-vec using `client.tool_runtime.rag_tool.insert()`.
+in batches of 100 into both Faiss and sqlite-vec.
 
 ```{image} ../../../../_static/providers/vector_io/write_time_comparison_sqlite-vec-faiss.png
 :alt: Comparison of SQLite-Vec and Faiss write times