eth: handle unresponsive rpcs

[nick1udwig]

Problem

1. ETH runtime module is brittle to RPC failure
2. hypermap-cacher is too closely built around Alchemy API settings (which just changed!)

Solution

1. Improve ETH runtime module by making RPC entries definitely ordered and by tracking what methods an RPC has failed to serve (and skipping that RPC for those methods in future; and retrying on RPC failures to allow for un-marking of failure if it is intermittent)
2. Reduce that. Instead we now try to getLogs for the entire history of logs and halve the log range until it works

Testing

• Getting logs from scratch without any hypermap-cacher bootstrapping nodes
• Falling back from one RPC to another
• Proper setting of log range for range-limited RPC providers

Docs Update

None

Notes

Future work on/around ETH runtime module:

1. Script(s) to change ordering of ETH RPCs

@dolled-possum