perf(ai-gateway): read provider index from replica

[chrarnoldus]

Summary

• read the cached models-by-provider snapshot from readDb instead of the primary
• keep strongly consistent gateway reads, including balances, rate limits, BYOK credentials, and experiment routing, on the writer

Rationale

The provider index already has a 30-second in-process cache and has no read-after-write dependency, so it can tolerate normal replica lag while reducing primary traffic before OpenRouter requests.

Validation

• pnpm exec oxfmt apps/web/src/lib/ai-gateway/providers/openrouter/models-by-provider-index.server.ts
• pnpm exec oxlint --config .oxlintrc.json apps/web/src/lib/ai-gateway/providers/openrouter/models-by-provider-index.server.ts
• pnpm --filter web test --runInBand src/app/api/organizations/[id]/defaults/route.test.ts src/routers/model-preferences-router.test.ts src/routers/organizations/organization-settings-router.test.ts
• pnpm --filter web typecheck
• git diff --check

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

Single-line data-source swap (db → readDb) for a cache-backed, read-after-write-independent query; no issues found in the changed lines.

Files Reviewed (1 files)

• apps/web/src/lib/ai-gateway/providers/openrouter/models-by-provider-index.server.ts

---

_{Reviewed by claude-sonnet-5 · Input: 20 · Output: 3.6K · Cached: 457.6K}

_{Review guidance: REVIEW.md from base branch main}