Support catalog messages for non-request approval policies (#34434)

## What changed

- Add model-catalog approval message variants for `never` and `unless_trusted`.
- Select the catalog message that matches the active approval policy, while retaining the existing built-in text when that variant is absent.
- Treat an explicitly empty variant as an instruction to suppress the built-in approval text, consistent with `on_request` messages.

## Testing

- Cover variant selection, fallback and empty-message behavior, catalog deserialization, and the initial permissions message sent to the model.

GitOrigin-RevId: a0f8d41a08645f39b80093be53f200eeee18ca25

codex-rs/protocol/src/openai_models.rs

@@ -515,6 +515,8 @@ pub struct ModelMessages {
 pub struct ApprovalMessages {
     pub on_request: Option<String>,
     pub on_request_auto_review: Option<String>,
+    pub never: Option<String>,
+    pub unless_trusted: Option<String>,
 }
 
 #[derive(Debug, Serialize, Deserialize, Clone, PartialEq, Eq, TS, JsonSchema)]
@@ -765,7 +767,8 @@ mod tests {
                 "instructions_template": null,
                 "instructions_variables": null,
                 "approvals": {
-                    "on_request": ""
+                    "on_request": "",
+                    "never": ""
                 }
             }"#,
         )
@@ -776,6 +779,8 @@ mod tests {
             Some(ApprovalMessages {
                 on_request: Some(String::new()),
                 on_request_auto_review: None,
+                never: Some(String::new()),
+                unless_trusted: None,
             })
         );
     }