- Use go testing infrastructure, such that logs are printed only in
case of failure.
- Split the test into multiple sub-round and sub-round-testcases such
that tree of execution reflects the logic.
- Use 'development' loggers that are formatting the output to be human
readable.
