fix(#229): Remediate code review findings for performance tests

- Fix CRITICAL: Increase single-spawn threshold from 10ms to 50ms (CI flakiness)
- Fix CRITICAL: Replace no-op validation test with real backoff scale tests
- Fix IMPORTANT: Add warmup iterations before all timed measurements
- Fix IMPORTANT: Increase scan position ratio tolerance to 10x for sub-ms noise
- Refactored queue perf tests to use actual service methods (calculateBackoffDelay)
- Helper function to reduce spawn request duplication

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

apps/orchestrator/tests/performance/secret-scanner-throughput.perf-spec.ts

@@ -115,9 +115,9 @@ describe("Performance: Secret Scanner", () => {
       expect(duration1).toBeLessThan(100);
       expect(duration2).toBeLessThan(100);
 
-      // And be within 5x of each other (no pathological behavior)
-      const ratio = Math.max(duration1, duration2) / Math.max(0.01, Math.min(duration1, duration2));
-      expect(ratio).toBeLessThan(5);
+      // Both should complete within a reasonable ratio (allowing for sub-ms noise)
+      const ratio = Math.max(duration1, duration2) / Math.max(0.1, Math.min(duration1, duration2));
+      expect(ratio).toBeLessThan(10);
     });
   });
 });