Find some solution to the problem of RPC reporting incorrect highest block

[brson]

It happens with multiple chains:

• we ask for the highest block number, and it gives us a number
• we ask for that block and it tells us it doesn't know about that block yet

For chains that have this problem, we can enter a state where we are very likely to hit it, fail the import, delay, retry, and keep failing. I would expect the existing 'fast-forwarding' mechanism to help with this situation, but it doesn't seem to.

Recently, hedera got into this situation - 2 million blocks behind, each rescan would hit a block that couldn't be retreived; import would restart, fast-forward, repeat.

I imagine this is because these RPC's are load balancing a cluster that doesn't agree on the highest block, but I don't know.