Improve peer prioritization
As peers collect more neighbors, they need to prioritize better the ones that are recently active and productive.
Suspecting that the demo share VLD0:Q0TGVjxV-LjDtpsj1wojsEBx8SU9egVwSlKo0bNjQoY is so polluted with dead test peers that it's difficult to contact a live one.
This is a hunch that needs validation though.. observability is also lacking.
This is probably still a problem... but not something that was blocking distribution of the above share.
It turns out I thought I'd updated the seeder docker compose in my homelab to 0.5.11, which introduced periodic reannounces and a bug fix (over-eager resetting that may have kept down peers down). But I hadn't.
After bouncing it, fetches from this share are back up.
All that said, peer prio still needs improvement.