perf(api): high baseline memory usage across API endpoints (RSS >512MB, heap >256MB)

Problem

Clinic doctor profiling reveals elevated memory usage across the majority of API endpoints. Out of 28 endpoints profiled, 17 exceeded the 512MB RSS threshold and 14 exceeded the 256MB heap threshold under 10 concurrent connections for 10s.

This is a systemic issue — not specific to any one endpoint — suggesting the server's baseline memory footprint is high before request handling even begins.

Worst offenders (RSS max)

Investigation areas

• Startup allocations: What is the RSS/heap immediately after server boot, before any requests?
• Connection pools: Are DB connection pools (Sequelize + Drizzle) over-allocated?
• Cached objects: In-memory caches (Memoize decorators) may hold large provider/GPU lists that are never freed
• Module loading: The bundled rest-app.js is 338KB — check if eager module initialization allocates large objects

Suggested next steps

1. Profile server at idle (no requests) to establish baseline memory
2. Check @Memoize decorated methods for unbounded cache growth
3. Review connection pool sizes for both Sequelize and Drizzle
4. Consider lazy initialization for large data structures

Context

Part of the API event loop performance audit. Full results in apps/api/profile-results.json.

Start with apps/api/profile-results.json and profile the server at idle to establish its RSS and heap baseline. Then inspect @Memoize methods, Sequelize and Drizzle pool settings, and eager initialization associated with rest-app.js. Done means the baseline and the principal source of the excessive memory usage are established with profiling evidence.