Have a persistent list of interesting events, ideally with an attached RSS feed. Things like
- saw a bad block
- orphaned block/reorg
- saw a bad transaction
- large mempool eviction
- etc.
This will be somewhat disparate from alertmanager in that alertman surfaces a lot of stuff that isn't "interesting" for end consumers of bmon; e.g. node is low on disk space or is down.