Rebalance Front conversations off agents who go offline or over capacity

Runs every few minutes to catch conversations stranded on agents who have clocked out or blown past their personal load cap. It pulls those conversations and redistributes them to teammates who are on shift and under cap, then posts a short summary to Slack so the team sees what moved.

Use it when agents end shifts mid-conversation or get buried during a spike, leaving tickets sitting unanswered on someone who is gone. This keeps the queue flowing without a supervisor babysitting it.

1. 1A schedule (e.g. every 5 minutes) triggers the run.
2. 2The flow reads the agent roster and per-agent caps from Postgres and pulls all open assigned conversations from Front.
3. 3A logic step flags conversations owned by off-shift or over-cap agents.
4. 4For each flagged conversation it selects an on-shift, under-cap recipient by lowest current load.
5. 5It reassigns those conversations in Front and records each move in the Postgres audit log.
6. 6A Slack message summarizes how many conversations moved and to whom.