[100K DAU][M2] Run systematic ghz baselines at varying concurrency

[api-Hypernova]

Context

Part of M2: Test Infrastructure & Baseline under epic #1834.

Task

Run ghz benchmarks for each relevant D14N API at concurrency levels: 1, 10, 50, 100, 200+ workers. Record RPS, avg latency, P99, error rate.

Acceptance Criteria

• Baselines for all read APIs (QueryEnvelopes, GetInboxIds, GetNewestEnvelope)
• Baselines for all write APIs (Welcome, GroupMessage-256B, GroupMessage-5KB, PublishClientEnvelopes)
• Results at each concurrency level documented
• Per-API scaling curves (RPS vs concurrency vs latency) graphed

Estimated Duration

2-3 days