As of WT-16104, WiredTiger skips checkpoints entirely when it is running as a standby. Local tables continue to work, albeit without any consistency guarantees about their contents when the connection gets closed.

We have two options apart from keeping the status quo:

1. Perform checkpoints on the standby unconditionally. This has the benefit of fully supporting local tables, but at the cost of reintroducing more cases to consider when debugging checkpoint failures, especially race conditions (eliminating them was the motivation behind WT-16104).
2. Perform checkpoints on the standby if configured to do so, such as if the connection-level parameter lose_all_my_data is set to false. This will have the benefit of not having to consider these extra cases for most failures outside of WiredTiger testing, while providing the durability guarantees required by our testing.

Please note that in addition to enabling these checkpoints, we should bring back the step-up race condition test that was removed as a part of WT-16104.