This issue was discovered in WT-10862, which was investigating the impacts of introducing support to read fast-truncate information in versions 6.0 and below.
If there is a stable fast-truncated tree in a newer version and we downgrade to the older version 6.0, we may leak disk space as we ignore the page deleted information in version 6.0. Checkpoint cleanup will ignore the fast-deleted pages, leading to a disk space leak.
We need to ensure that checkpoint cleanup in older versions is correctly handling fast-deleted pages before we start allowing MongoDB to write fast-truncate information.