readinessProbe/livenessProbe hit /metrics on top of Prometheus's own
scrapes; exporter re-logs into actual_server on every call (no session
reuse), tripped actual_server's login rate limit (too-many-requests).
Switch probes to tcpSocket and stretch scrape interval to 5m.
Prometheus exporter for Actual Budget. actual_server itself stays in
homelab-docker-compose-prod (host docker-compose), exporter only scrapes it
over LAN and exposes /metrics via a ServiceMonitor. No HTTPRoute. Password
and budget ID come from a hand-created Secret, not committed.