Scheduled autonomy

The loop is driven by time-triggered cron jobs (e.g. ~15-minute Spark telemetry refreshes and the self-improve loop) so the agent acts on a fixed cadence rather than only on demand.

grounded in: self-improve-loop.sh plus the logs/crontab.backup.* snapshots and the ~15-min cadence of 'spark: 2-node GB10 telemetry refresh' commits