[memory refactor][5/n] Migrate all vector_io providers (#835)

See https://github.com/meta-llama/llama-stack/issues/827 for the broader
design.

This PR finishes off all the stragglers and migrates everything to the
new naming.

llama_stack/providers/tests/tools/fixtures.py

@@ -83,7 +83,7 @@ async def tools_stack(
 
     providers = {}
     provider_data = {}
-    for key in ["inference", "memory", "tool_runtime"]:
+    for key in ["inference", "vector_io", "tool_runtime"]:
         fixture = request.getfixturevalue(f"{key}_{fixture_dict[key]}")
         providers[key] = fixture.providers
         if key == "inference":
@@ -117,7 +117,12 @@ async def tools_stack(
     )
 
     test_stack = await construct_stack_for_test(
-        [Api.tool_groups, Api.inference, Api.memory, Api.tool_runtime],
+        [
+            Api.tool_groups,
+            Api.inference,
+            Api.vector_io,
+            Api.tool_runtime,
+        ],
         providers,
         provider_data,
         models=models,