use a more lightweight probe most of the time (i am thinking using FindHeaders) to reduce network load and local CPU load

[softminus]

only run the block check occasionally (like once every two hours on average for each node), since it requires transferring lots of data and hashing locally

to serve a node, require that it has served a valid block (hashes OK) recently enough and that none of the information in the handshake has changed since then.

use a lightweight probe for the uptime probing: either FindHeaders (which would require doing a bit of hardcoding and would mean we can't use the zebra checkpoints, ah well) or just a protocol negotiation