Bounce-Rate Spike Sentinel: Pause Sends by Domain and Alert Owner

Every 30 minutes this workflow reads send and bounce counts from your email event table, computes the hard-bounce rate per recipient domain over a rolling window, and isolates any domain that breaches your ceiling (default 5%). For each breached domain it flips a `paused` flag in your sends table so no further campaigns go out to that domain, records an incident row, and notifies the responsible list owner in Slack with the exact bounce rate, volume, and top bouncing addresses.

Run this when a misconfigured DNS record, a stale import, or a spam-trap-laden list can quietly torch your sender reputation at one mailbox provider before a human notices. It contains the blast radius to a single domain instead of pausing your whole program.

1. 1A schedule fires every 30 minutes.
2. 2Postgres returns sends and hard bounces grouped by recipient domain for the rolling window.
3. 3A logic step keeps only domains above the bounce-rate threshold with enough volume to matter.
4. 4Postgres sets `paused = true` for queued sends to each breached domain and inserts an incident record.
5. 5Slack posts the domain, bounce rate, sample addresses, and a link to the incident to the owner channel.