Fix slow ORM vs native pg - runtime optimizer or drop-in proxy

Executive Summary

Developers report Prisma ORM 10x-30x slower than native pg on Node-alpine with pg-bouncer, hurting read and write latency. Build a drop-in runtime optimizer, lightweight native bindings, or proxy that matches native pg latency while preserving ORM ergonomics. Developers are deploying nodejs server images like Node-alpine with pg-bouncer widely, and observability tools now surface that ORM runtime overhead, not the database, is the bottleneck - the source slack thread and tracing reported this exact pattern. Serverless and per-request billing make latency and resource cost visible on monthly bills, increasing willingness to pay for runtime optimization. Recent tooling advances - safe Rust native modules and improved node native addon ecosystems - make shipping a high-performance, cross-platform runtime binding or proxy practical now. Offer a drop-in optimizer or proxy that reproduces native pg latency for Prisma users by combining low-level language bindings (Rust or native node addons), query batching and prepared statement reuse, and targeted instrumentation. Use telemetry from real production Prisma stacks to build a performance dataset - this creates a feedback moat because we can tune heuristics for typical Node-alpine plus pg-bouncer deployments and surface per-call fixes automatically. The source shows developers already trace the problem in production and want a direct fix - a drop-in proxy or native binding is the fastest route to adoption.