Keep Guardian v2 risk scores in memory (#39304)

## What changed

- Stop writing Guardian v2 security risk scores to rollout history.
- Treat resumed and forked threads as having no previous Guardian score, so their
  first tool approval is classified and reviewed normally.
- Remove the now-unused `codex-history` dependency from the Guardian v2
  extension.

## Testing

- Update app-server coverage for new, resumed, and forked threads to wait for
  classification completion and verify the expected approval reviews.

GitOrigin-RevId: 74d9caa903a1ff9d38e577b4697b5f8970f11ebb

codex-rs/ext/guardian-v2/Cargo.toml

@@ -17,7 +17,6 @@ codex-api = { workspace = true }
 codex-core = { workspace = true }
 codex-extension-api = { workspace = true }
 codex-features = { workspace = true }
-codex-history = { workspace = true }
 codex-http-client = { workspace = true }
 codex-login = { workspace = true }
 codex-model-provider = { workspace = true }

codex-rs/ext/guardian-v2/src/extension.rs

@@ -27,7 +27,6 @@ use codex_extension_api::ToolName;
 use codex_extension_api::ToolPayload;
 use codex_extension_api::ToolStartInput;
 use codex_features::Feature;
-use codex_history::RolloutItem;
 use codex_login::AgentIdentityAuthPolicy;
 use codex_login::AuthManager;
 use codex_model_provider::create_model_provider;
@@ -614,12 +613,6 @@ impl ToolLifecycleContributor for GuardianV2Extension {
                     .latest_scored_tool_call
                     .fetch_max(tool_call_index, Ordering::Release);
                 classification_finished_at = Some(Instant::now());
-                if !config.ephemeral {
-                    thread
-                        .append_rollout_items(&[RolloutItem::SecurityRiskScore(score)])
-                        .await
-                        .map_err(|error| error.to_string())?;
-                }
                 Ok("success")
             }
             .await;

codex-rs/ext/guardian-v2/src/extension_tests.rs

@@ -18,7 +18,6 @@ use codex_extension_api::ToolName;
 use codex_extension_api::ToolPayload;
 use codex_extension_api::ToolStartInput;
 use codex_features::Feature;
-use codex_history::RolloutItem;
 use codex_login::AuthManager;
 use codex_login::CodexAuth;
 use codex_login::ExternalAuth;
@@ -1201,21 +1200,6 @@ async fn contributor_samples_tool_calls_with_the_existing_luna_pool() -> Result<
             .await,
         None
     );
-    test.codex.ensure_rollout_materialized().await;
-    test.codex.flush_rollout().await?;
-    let persisted_scores = test
-        .codex
-        .load_history(/*include_archived*/ false)
-        .await?
-        .items
-        .into_iter()
-        .filter_map(|item| match item {
-            RolloutItem::SecurityRiskScore(score) => Some(score),
-            _ => None,
-        })
-        .collect::<Vec<_>>();
-    assert_eq!(persisted_scores, vec![score.as_ref().clone()]);
-
     thread_store.insert(SecurityRiskScore {
         scores: BTreeMap::from([("action_risk".to_string(), 0.5)]),
         sampled_at: None,