Clarify cost KPI: name production topology + 15-min review framing

[talsyg]

Summary

The exercise asks candidates to optimize for throughput and cost, but cost was under-specified. Candidates would default to scaling answers (which is throughput) because nothing in the README signaled the cost shape of the infra. We were waiting for answers like batching to the queue (per-request pricing) and reducing cross-AZ DB round-trips, but the README didn't make those levers visible.

This PR makes two changes to the README so the existing prompt does the work, without spoon-feeding answers:

• Add a "Production Deployment" section naming the real infra: 2× EC2, SQS, RDS Postgres in a different AZ. A senior reading "SQS + cross-AZ RDS" should immediately recognize per-request queue pricing and per-INSERT cross-AZ cost — without us writing a cost-model paragraph that gives the answers away.
• Reframe "Your Task" to match how the exercise is actually run: a 15-minute code review producing the top 3 changes, then prompted to an AI assistant. Anchors the 3 picks to "the production topology above" so ignoring infra context is an explicit miss.

Also includes the prior commit on this branch (architecture diagram upgrade).

Test plan

• Read the new "Production Deployment" section — topology is clear and matches reality
• Confirm "Your Task" still reads as open-ended (no hint sheet, no rubric)
• Sanity-check that local Quick Start still works as-is (no infra changes, only README)

🤖 Generated with Claude Code

[entro-pr-scanner]

🔐 Entro Security Scan Results

✅ No Secrets Found

Great news! No secrets were detected in PR #2.

---

ℹ️ About Entro Security

Entro is The Enterprise Security Platform for AI Agents & Non-Human Identities that helps organizations discover, monitor, and protect their secrets across the entire development lifecycle. This scan automatically detects exposed credentials, API keys, tokens, prompt-injection attempts, and other sensitive data in your code to prevent security breaches before they happen.