Roadmap: bigger swings

[obeone]

Bigger-swing ideas that came up during audits but are well past the scope of any single PR. Tracking them here as one list instead of opening ten near-empty issues.

• Persistent hub state (survive a restart instead of losing peers and log)
• Multi-hub federation
• LLM-generated backlog digest for a peer that's been away
• A semantic "claim" primitive (claim a task or resource, not just message about it)
• Prometheus metrics
• Replay / time-travel console for past sessions
• Read receipts
• Room templates
• Webhook/Slack egress
• Connector registry