perf(web): parallelize homepage data loading

[jonathanhaaswriter]

Summary

• keep the existing dashboard cache warm lifecycle while removing the first warm from the server-readiness critical path
• start independent dashboard, readiness, and coverage reads together so secondary data does not form a client waterfall
• make the scale benchmark wait for real homepage work data and serve production-shaped dashboard fixtures
• cover cold/warm cache behavior, tenant and workspace separation, non-blocking secondary stalls, and non-blocking startup warm behavior

Validation

• NODE_OPTIONS=--localstorage-file=/tmp/cerebro-dashboard-architecture-vitest-localstorage npm test -- --run (114 files, 693 tests)
• npx eslint src/instrumentation.ts src/instrumentation.test.ts src/app/page.tsx src/app/page.test.tsx 'src/app/api/cerebro/[...path]/route.test.ts' scripts/grc-scale-benchmark.mjs
• npm run build
• CEREBRO_GRC_SCALE_ROUTE=home CEREBRO_GRC_SCALE_USABLE_MS=1000 node scripts/grc-scale-benchmark.mjs --quick --production --artifacts (Home usable 951ms, API 32ms, 292 DOM nodes)