feat: add runnable benchmark suite for diff viewer comparison

[Zhang-JiahangH]

Motivation

• Provide a reproducible, runnable benchmark harness to compare rendering performance of react-virtualized-diff against common alternatives (react-diff-viewer, react-diff-view) on large inputs prior to release.
• Measure practical metrics for decision-making: initial render latency, average FPS during an auto-scroll scenario, and JS heap usage for a variety of dataset sizes.

Description

• Added a Vite benchmark app at apps/benchmark which generates synthetic diffs and can render any of the three libraries via URL params; new files include apps/benchmark/src/main.tsx, apps/benchmark/index.html, apps/benchmark/package.json, apps/benchmark/tsconfig.json, and apps/benchmark/vite.config.ts.
• Added an automated runner scripts/run-benchmark.mjs that launches the benchmark app and uses Playwright to exercise every combination of library and sizes (1k / 10k / 50k / 100k lines), then writes results to benchmark-results/results.json and benchmark-results/results.md.
• Updated root package.json to add a benchmark script (node ./scripts/run-benchmark.mjs) and declared playwright as a dev dependency, and added README documentation describing how to run the suite and the produced output files.

Testing

• node --check scripts/run-benchmark.mjs passed static Node syntax check successfully.
• pnpm install failed in this execution environment due to npm registry authorization (403) so dependencies could not be installed and integration run could not be performed here.
• pnpm -r typecheck failed because apps/benchmark type dependencies (vite/client) were not available due to the failed install (TypeScript error TS2688).

---

Codex Task