After merging WT-14697, we regularly started seeing format-failure-configs-test failing. I spent some time diagnosing this and narrowed it down to test/format enabling precise checkpoints with disagg storage, which resulted in the (on-disk) block manager corrupting extent lists.

For the time being, I've gated test/format using precise checkpoint on disagg being enabled, but ideally we wouldn't need this restriction.

The definition of done here is to root-cause the corruption, and fix if applicable.

It's possible (but seems unlikely to me) there is a very good reason not to enable precise checkpoints with the old block manager. In this case we should more broadly review the precise checkpoint config, and perhaps tie it to disagg (making it an error to attempt to use precise checkpoint without disagg).