Automod rules and banned-word filters

## Summary
Implement automod rules and banned-word filtering with Discord-like moderator outcomes and self-hosting-friendly controls.

## Scope
- Define rule types for banned terms and other core content-matching triggers.
- Support moderator-configurable actions such as block, flag, or rate-limit escalation.
- Expose clear moderator controls, rule ordering, and testability.
- Keep false-positive handling and user feedback understandable.

## Acceptance Criteria
- [ ] Moderators can create, edit, enable, disable, and delete automod rules from the product UI.
- [ ] Matching content is handled server-side before it becomes visible beyond intended policy.
- [ ] Users receive understandable feedback when automod blocks or alters their action.
- [ ] Moderators can inspect which rule fired and why.

## Operational Notes
- Self-hosted communities need controls that are powerful without becoming opaque policy code.
- Rule execution must be deterministic and observable under load.