This task focuses on isolating whether the performance degradation observed in GridScrollBenchmark.spec.mjs on larger viewports (Desktop) is an artifact of the Playwright test harness interacting with MutationObserver or the specific manual scroll implementation.

Objectives

1. Create Benchmark Variant: Implement test/playwright/e2e/GridNativeScrollBenchmark.spec.mjs.
2. Disable Observers: Explicitly disconnect Neo.main.DomAccess.documentMutationObserver within the test context to eliminate observer overhead.
3. Use Native Scrolling: Replace the JS-driven setInterval loop with element.scrollTo({ behavior: 'smooth' }) to leverage the browser's compositor thread.
4. Verify: Run the benchmark across Mobile, Laptop, and Desktop viewports to see if the "Desktop Cliff" disappears.

Acceptance Criteria

• New benchmark file created and functional.
• Benchmark run successfully in headless mode.
• Results compared against the baseline GridScrollBenchmark.spec.mjs.