Restrict agent roles to bounded configuration overrides (#39299)

## Why

Agent roles should customize a child agent without expanding the authority or changing the provider configuration inherited from its parent session.

## What changed

- Apply only supported role overrides for model behavior, developer instructions, personality, service tier, and capability reductions.
- Preserve parent-owned permissions, model providers, endpoints, MCP servers, notifications, and other unrestricted configuration.
- Keep managed feature requirements effective when a role disables capabilities, and reject symlinked user role files.
- Use the same bounded role application path for both multi-agent implementations and resumed agents.

## Testing

Add coverage for authority preservation, managed feature requirements, symlink rejection, provider inheritance, and provider routing after resume.

GitOrigin-RevId: c528d615b691f9c02bfbc21154d514ea07743010

codex-rs/core/src/tools/handlers/multi_agents_common.rs

@@ -1,5 +1,4 @@
 use crate::agent::role::apply_role_to_config;
-use crate::agent::role::apply_role_to_config_for_multi_agent_v2;
 use crate::config::Config;
 use crate::config::DEFAULT_MULTI_AGENT_V2_MIN_WAIT_TIMEOUT_MS;
 use crate::config::HARD_MAX_MULTI_AGENT_V2_TIMEOUT_MS;
@@ -389,15 +388,9 @@ pub(crate) async fn apply_spawn_agent_role(
 ) -> Result<(), FunctionCallError> {
     let previous_model = config.model.clone();
     let previous_reasoning_effort = config.model_reasoning_effort.clone();
-    if session.multi_agent_version() == Some(MultiAgentVersion::V2) {
-        apply_role_to_config_for_multi_agent_v2(config, role_name)
-            .await
-            .map_err(FunctionCallError::RespondToModel)?;
-    } else {
-        apply_role_to_config(config, role_name)
-            .await
-            .map_err(FunctionCallError::RespondToModel)?;
-    }
+    apply_role_to_config(config, role_name)
+        .await
+        .map_err(FunctionCallError::RespondToModel)?;
     if config.model == previous_model && config.model_reasoning_effort == previous_reasoning_effort
     {
         return Ok(());

codex-rs/core/src/tools/handlers/multi_agents_tests.rs

@@ -919,6 +919,7 @@ async fn multi_agent_v2_spawn_partial_fork_turns_allows_agent_type_override() {
         multi_agent_version: codex_protocol::protocol::MultiAgentVersion::V2,
         ..turn
     };
+    let parent_provider_id = turn.config.model_provider_id.clone();
 
     let output = SpawnAgentHandlerV2::default()
         .handle(invocation(
@@ -952,7 +953,7 @@ async fn multi_agent_v2_spawn_partial_fork_turns_allows_agent_type_override() {
         .await;
 
     assert_eq!(snapshot.model, "gpt-5-role-override");
-    assert_eq!(snapshot.model_provider_id, "ollama");
+    assert_eq!(snapshot.model_provider_id, parent_provider_id);
     assert_eq!(snapshot.reasoning_effort, Some(ReasoningEffort::Minimal));
 }