Don't block bonder operations if the redundant RPC is down

[shanefontaine]

Note

Consider a world where the redundant RPC is removed. It can be replaced by the current RPC endpoint provided and still achieve the same goal. Maybe the bonder can bring as many RPCs as they want and the aggregate is used?

In the current implementation, the redundant RPC blocks transactions if it is down. Seeing as it is difficult to run even a single RPC, this should not happen.

Possible implementation:

• By default, update the redundant verification to use the normal endpoint. If a redundant one is passed in, use that only.
  • Don’t fallback is sensitive since that defeats the purpose
  • Redundant shouldn’t block anything