Introduce a "Router" layer for providers

Some providers need to be factorized and considered as thin routing
layers on top of other providers. Consider two examples:

- The inference API should be a routing layer over inference providers,
  routed using the "model" key
- The memory banks API is another instance where various memory bank
  types will be provided by independent providers (e.g., a vector store
  is served by Chroma while a keyvalue memory can be served by Redis or
  PGVector)

This commit introduces a generalized routing layer for this purpose.

llama_toolchain/memory/providers.py

@@ -26,16 +26,16 @@ def available_providers() -> List[ProviderSpec]:
             config_class="llama_toolchain.memory.meta_reference.faiss.FaissImplConfig",
         ),
         remote_provider_spec(
-            api=Api.memory,
-            adapter=AdapterSpec(
+            Api.memory,
+            AdapterSpec(
                 adapter_id="chromadb",
                 pip_packages=EMBEDDING_DEPS + ["chromadb-client"],
                 module="llama_toolchain.memory.adapters.chroma",
             ),
         ),
         remote_provider_spec(
-            api=Api.memory,
-            adapter=AdapterSpec(
+            Api.memory,
+            AdapterSpec(
                 adapter_id="pgvector",
                 pip_packages=EMBEDDING_DEPS + ["psycopg2-binary"],
                 module="llama_toolchain.memory.adapters.pgvector",