Measure the complete install-to-first-window developer loop

[OthmaneBlial]

Context

The public benchmark covers Research Desk runtime and package metrics, while the roadmap also asks for install, first compile, warm start, frontend HMR, and package-generation measurements. Those numbers need versioned receipts and explicit host context.

Affected area

• benchmark scripts and receipts
• docs/developer-loop.md
• site/benchmarks.html
• CLI phase and timing output

Acceptance criteria

• Measure CLI install, new project creation, cold first native build, warm restart, one frontend-only edit, and package generation.
• Store raw samples, median, sample count, host, toolchain, source commit, and cache state in a versioned JSON receipt.
• Fail the generator when required fields or source identity are missing.
• Display limitations next to every public metric.
• Add a documented refresh procedure that does not overwrite historical receipts.

Test command

node scripts/benchmark_developer_loop.mjs --help
node scripts/check_site.mjs
npm --prefix site test -- tests/site.spec.mjs

Non-goals

• A universal speed promise.
• Comparing against competitors without equivalent clean-host methodology.
• Hiding Rust or frontend dependency installation time.