perf(api): high loop utilization on multiple endpoints (clinic event-loop category)

Problem

Clinic doctor flagged 7 endpoints with loopUtilization=performance, categorizing them as event-loop bottlenecked. High loop utilization (>90%) means the event loop is saturated and has little headroom for additional work.

Affected Endpoints

Analysis

While event loop delay is low for these endpoints, the loop utilization is near 100%. This means:

• The server has no spare capacity during sustained load
• Any additional CPU work (other requests, timers, garbage collection) will immediately cause delays
• Under production traffic with mixed endpoints, these will degrade faster than profiled in isolation

Suggested investigation

• These endpoints likely do synchronous JSON serialization or computation that keeps the loop busy
• Consider response caching at the HTTP layer (ETag/304) to avoid re-serialization
• For /v1/graph-data/compute, check if the map/filter/reduce chains can be pre-computed
• For /v1/templates-list, check JSON.parse/stringify overhead

Context

Part of the API event loop performance audit.

Start by profiling the seven listed API endpoints, especially /v1/graph-data/compute and /v1/templates-list, under sustained load. Check synchronous serialization, JSON.parse/stringify overhead, and the map/filter/reduce work described in the investigation notes. Done means the event-loop audit shows reduced utilization and the endpoints retain their expected behavior.