solutions-engineering: tune lab bottleneck signals

[adityamaru]

Summary

• Make the fraud benchmark start from a 4-core runner label and increase its fixed CPU workload so it is the obvious first performance bottleneck.
• Run both Docker platform builds on an x64 runner label so the arm64 target exercises cross-platform build behavior.
• Make the Dockerfile intentionally less cache-friendly and add native build work so Docker build optimization has meaningful surface area.

Test plan

• Ran npm run lint in solutions-engineering/migration-lab/app.
• Ran DOCKER_BUILD_TASKS=4 DOCKER_BUILD_ITERATIONS=500000 DOCKER_BUILD_WORKERS=2 node scripts/docker-build-workload.js.
• Ran PROFILE_CPU_TASKS=4 PROFILE_CPU_ITERATIONS=500000 PROFILE_CPU_WORKERS=2 npm run profile:cpu.

Made with Cursor

---

^{Need help on this PR? Tag @codesmith with what you need.}

• Let Codesmith autofix CI failures and bot reviews

---

^{Need help on this PR? Tag @codesmith with what you need.}

• Let Codesmith autofix CI failures and bot reviews (Staging)