fix(ai): separated codex orchestration usage

- Added a Usage.orchestration sidecar for provider-side service tokens so Responses/Codex totals and costs stay accurate without inflating visible prompt input/cache buckets.
- Updated Codex/WebSocket usage, session/status aggregates, and usage reporting to preserve orchestration-aware totals.
- Added regressions for OpenAI Responses accounting, Codex WebSocket terminal usage, cost calculation, and session aggregation.

Fixes #4469

packages/coding-agent/test/core/js-workflow-helpers.test.ts

@@ -89,6 +89,10 @@ describe("executeJs workflow helpers", () => {
 				output: 777,
 				cacheRead: 0,
 				cacheWrite: 0,
+				totalTokens: 787,
+				orchestrationInput: 0,
+				orchestrationOutput: 0,
+				orchestrationCacheRead: 0,
 				premiumRequests: 0,
 				cost: 0,
 			}),