Improve retry handling in yahcli

Yahcli uses a naive retry approach for failed requests: repeatedly send the same request to the same node X number of times. We want to improve this strategy by implementing two simple changes: first, by sending the request to other healthy nodes in the configured network; second, by adding a wait time (e.g. 1-5 seconds) in between retry requests.

Start by locating yahcli's failed-request retry entry point and the existing logic for configured healthy nodes. Confirm how retry counts are currently handled, then verify that retries can use other healthy nodes and include a 1–5 second wait between requests; add or update tests for both behaviors.

Written by the indexing model from the issue text.