fix(scaletozero): skip scale-to-zero toggle for loopback connections

[tnsardesai]

Summary

Loopback requests — such as internal health checks / playwright daemon connections — were toggling the scale-to-zero control file, keeping VMs alive indefinitely and driving up usage costs.

Changes:

• Middleware now detects loopback source addresses and passes them through without touching scale-to-zero state
• Added info-level logging around control-file writes for better observability
• Added unit tests for the middleware and the loopback detection helper

---

Note

Low Risk
Small, localized change to request middleware behavior plus additional logging and tests; primary risk is mis-detecting client IPs when RemoteAddr doesn’t reflect the true origin behind proxies.

Overview
Prevents internal/loopback HTTP traffic from affecting VM scale-to-zero behavior by having the scaletozero HTTP middleware bypass Disable/Enable when RemoteAddr is a loopback IP (via new isLoopbackAddr).

Adds unit coverage for the middleware behavior (external vs loopback, and disable error handling) and increases controller observability by logging info-level messages when the scale-to-zero control file is missing or successfully written.

^{Written by Cursor Bugbot for commit 8cee14f. This will update automatically on new commits. Configure here.}

[Sayan-]

yeet