[test][scheduler] Add the Scheduler benchmarks to test/benchmark #24137

Summary

Add the Scheduler to test/benchmark once #24028 lands, so its benchmarks run in a real browser with the benchmark CLI of @mui/internal-benchmark, compared against a baseline commit and reported on the dashboard.

Why

The Scheduler benchmarks (packages/x-scheduler-internals*/src/**/*.bench.ts) are pure-function microbenchmarks run with vitest bench in jsdom. They don't run in CI, they aren't compared against a baseline (the October comparisons in #23961, #23962, #23963, #23964 and #23984 were done by hand), and they don't measure what users actually experience: mounting a view, scrolling, dragging. A browser run with the production build is more representative, and test/benchmark gives us the baseline comparison and the dashboard for free.

Scope

• test/benchmark/src/scheduler/<Component>/ with scenarios.tsx + <Component>.bench.tsx, following the charts layout, and a src/smoke/scheduler.bench.tsx with one case per component for pull requests.
• Component scenarios we don't measure today: Event Calendar month view with 1,000 events (system and display timezone), Week view with recurring events, Event Timeline with 10,000 events and a scroll gesture, a drag interaction.
• Port the existing pure-function benches (day grid, time axis, dependency engine, store, recurring edits) to benchmark(), and remove bench:jsdom if nothing else uses it.
• Add @mui/x-scheduler, @mui/x-scheduler-premium and their peers to test/benchmark/package.json.
• Extend the CI job: include packages/x-scheduler* in the change gate, run scheduler/ on master, and run the scheduler files on pull requests that touch the scheduler packages.
• Event datasets generated inside the benchmarks (no imports from docs/), and a license key in the Premium pages so the watermark doesn't add a render.

Notes

• Blocked by [test][charts] Port the benchmarks to the benchmark CLI #24028.
• The Timeline is Premium, so check how the Charts Premium benchmarks handle the license key.