round robin or similar algo w/retries for rotating when we hit bad RPC clients

[pyramation]

sometimes a tx or query can fail, and we need to have a redundant way to make queries/txs

• [x] default endpoints
• [x] retry logic when failure is understood to be issue with a node/peer
• [x] allow user to add endpoints

related https://github.com/osmosis-labs/telescope/issues/14

[pyramation]

closing as this should be in something like cosmos kit and not telescope