retry with other rpcAddrs in event of RPC failure

[turadg]

What is the Problem Being Solved?

https://main.agoric.net/network-config lists multiple rpcAddrs. Currently most dapps take the first (such as agoric-cli does) or a random one (such as wallet-app does).

When an RPC server is down or impaired (such as in Agoric/dapp-inter#199 ), there's no way for the app to detect or recover.

Description of the Design

Provide a facility in ui-kit to abstract network-config for UIs.

Provide a way for the client to request a different rpcAddr than the last request. (Maybe the client keeps track of which it tried and can ask for "other than these" or maybe the ui-kit keeps track and client can say "different than I've had yet during this session")

Some examples or abstractions for RPC retrying using a different rpcAddr.

Security Considerations

The network-config list is powerful.

Scaling Considerations

The failures may be due to rate limiting so failures shouldn't be treated as bad server.
The client shouldn't hose the servers round-robin. Something low level could ensure a limit in request rate.

Test Plan

TBD