fix(backend): include subgraphs in cached graph retrieval

The cached graph function was missing include_subgraphs=True parameter which
is needed to construct full credentials input schema. This was causing
test_access_store_listing_graph to fail.

autogpt_platform/backend/backend/server/routers/cache.py

@@ -54,6 +54,7 @@ async def get_cached_graph(
         graph_id=graph_id,
         version=version,
         user_id=user_id,
+        include_subgraphs=True,  # needed to construct full credentials input schema
     )