docs: centralize performance guidance and benchmark data

[Mohamed Mansour (mohamedmansour)]

Summary

• add a data-driven homepage Benchmark Explorer backed only by a compact generated summary
• make lazy rendering, lazy hydration, interaction hydration, state projection, DOM strategy, and measurement guidance distinct across the developer docs and AI authoring reference
• move cross-framework benchmark ownership and raw evidence to microsoft/webui-benchmarks, remove the obsolete in-repo SSR showdown, and document the validated two-input summary export workflow

Benchmark publication

• raw SSR/browser captures, runners, and full evidence live in microsoft/webui-benchmarks; WebUI commits only docs/.webui-press/state/benchmark-summary.json
• the final summary contains 14 product rows and five independent metrics: single-core RPS, FCP, LCP, forced-GC JS heap, and renderer private bytes
• the paired captures match source commit 9131a650236479277aaf6d6a94c461fc66864449 and fixture digest 7dce3789006d6ed472e0deb6e020754d11577ec392cf832f4a9af2e85d39cbde
• the browser tier completed 98/98 samples (14 cases x 7 rounds); browser outcomes remain separate selectable metrics and never enter SSR scoring
• the combined validated exporter is in microsoft/webui-benchmarks#7, whose CI is green
• the full latest.json duplication and all local raw result files are removed from this repository

Validation

• pnpm --dir docs build
• cargo xtask check
• benchmark exporter focused tests: 59/59
• benchmark repository tests: 61/61 repository, 264/264 JavaScript, full Rust workspace