docs: Scaling & capacity-planning guide

[joelpeace48-cell]

Background & current state

Operators/SREs have no guidance on sizing the system for load. Scale features (replicas NEW-034, partitioning NEW-035, caching NEW-036, autoscaling NEW-051) and load tests (NEW-115) will exist, but no doc ties them to capacity recommendations.

Goal

Author docs/SCALING.md: capacity-planning guidance (replicas, DB sizing, cache, RPC throughput, pool sizing) backed by load-test numbers.

Outline

• Reference architecture for N concurrent users; component sizing; bottleneck analysis; HPA settings (NEW-051); DB/replica/cache guidance; RPC throughput + breaker; numbers from NEW-115/116.

Task breakdown

• Write docs/SCALING.md with concrete recommendations + load-test references.
• Include a sizing table per traffic tier.
• Cross-link from architecture + runbook.

Acceptance criteria

• Recommendations are backed by load-test numbers (reference load-tests/).

Verification

• Numbers cross-checked vs NEW-115 results; review.

Difficulty: medium · Effort: M · documentation + performance