Centralize Guardian classification of multi-agent tool calls (#57455)

## What changed

Add `GuardianScope::for_multi_agent_tool` and use it for both tool review requirements and asynchronous scoring. Classify `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent` as native tools when they match the configured namespace, accepting default namespace spellings when none is configured. Keep review enablement checks at each call site.

GitOrigin-RevId: 9180ccdc629532f6c171d899dc5881e15b447141

codex-rs/core/src/tools/multi_agent_tool.rs

@@ -4,6 +4,7 @@
 use crate::session::session::Session;
 use crate::tools::context::ToolInvocation;
 use crate::tools::registry::CoreToolRuntime;
+use codex_protocol::openai_models::GuardianScope;
 use codex_tools::JsonSchema;
 use codex_tools::ResponsesApiNamespace;
 use codex_tools::ResponsesApiNamespaceTool;
@@ -123,10 +124,8 @@ impl CoreToolRuntime for MultiAgentV2ToolOverrides {
             .config
             .features
             .enabled(codex_features::Feature::GuardianNativeMultiAgentHistory)
-            && matches!(
-                self.handler.tool_name().name.as_str(),
-                "spawn_agent" | "send_message" | "followup_task" | "interrupt_agent"
-            )
+            && GuardianScope::for_multi_agent_tool(&self.tool_name(), self.namespace.as_deref())
+                .is_some()
     }
 
     fn wait_until_ready<'a>(&'a self, session: &'a Arc<Session>) -> Option<BoxFuture<'a, ()>> {