IF: net_plugin lib<->head catchup transistion change

[heifner]

When lib is much closer to head, it will be very likely for a node to find its head is less than lib and think it needs to sync from a peer when in reality it is only a bit behind. Need a better heuristic for when to switch from in_sync<->head<->lib catchup modes.

This might be as simple as a hard-coded delta from peer head before assuming it is syncing (or should sync), but will require some testing.

[heifner]

• sync_manager::recv_handshakerecv_handshake - head < msg.lib is more likely to be true. The inverse is also more likely to be true. Both conditions sync 1 & sync 2 need to not trigger lib sync unless there is a distance between head<->lib of (latency + sync_diff_range). Where sync_diff_range could be hard-coded. But I believe setting it to half of p2p-keepalive-interval-ms (default 10s) is a better option. p2p-keepalive-interval-ms makes sense as a value to tie it to because this corresponds to how long you wait before pinging a node if you have not heard from it. The default would be 5 seconds (10 blocks).