Scan skill roots concurrently (#33369)

## What changed

- Scan up to eight skill roots concurrently instead of loading them one at a time.
- Share scan capacity across host-skill and plugin-skill loads to bound aggregate filesystem work.
- Merge completed scans in input order so root precedence and error ordering remain deterministic.
- Skip canonicalization and parsing work for empty roots.

## Testing

- Add coverage for out-of-order scan completion and shared-capacity blocking.

GitOrigin-RevId: 33a47aaa02aee7fe31a050bb04d51c1e2043d6ae

codex-rs/core-plugins/src/manager_tests.rs

@@ -1928,6 +1928,7 @@ async fn load_plugin_skills_dedupes_overlapping_manifest_roots() {
         /*restriction_product*/ None,
         &SkillConfigRules::default(),
         /*plugin_skill_snapshots*/ None,
+        Arc::new(Semaphore::new(MAX_CONCURRENT_ROOT_SCANS)),
     )
     .await;
 
@@ -5822,6 +5823,7 @@ async fn load_plugins_ignores_project_config_files() {
         /*plugin_skill_snapshots*/ None,
         Some(Product::Codex),
         /*remote_global_catalog_active*/ false,
+        Arc::new(Semaphore::new(MAX_CONCURRENT_ROOT_SCANS)),
     )
     .await;