fix(reasoning): preserve effort through Codex translations

Carry Claude reasoning_effort/reasoning into OpenAI Chat, map into
OpenAI Responses reasoning.effort, and keep request-level effort
(incl. xhigh) across tool-result turns instead of collapsing to high.

Co-authored-by: Cursor <cursoragent@cursor.com>

open-sse/services/provider.js

@@ -159,12 +159,10 @@ export function hasThinkingConfig(body) {
   return !!(body.reasoning_effort || body.thinking?.type === "enabled");
 }
 
-// Normalize thinking config based on last message role
-// - If lastMessage is not user → remove thinking config
-// - If lastMessage is user AND has thinking config → keep it (force enable)
+// Normalize provider-native thinking config based on last message role.
+// OpenAI reasoning_effort is request-level and must survive tool-result turns.
 export function normalizeThinkingConfig(body) {
   if (!isLastMessageFromUser(body)) {
-    delete body.reasoning_effort;
     delete body.thinking;
   }
   return body;