Checkpoint abandon is a dangerous operation that has the potential to lose data if PALI gets the checkpoint wrong, we'd like to have a more thorough defence which involves LS, specifically:

• When WT issues abandon, PALI queries the checkpoint manager to decide what is the phylog LSN which we believe we are abandoning after.
• When LS receives the abandon, it compares that phylog lsn to its own knowledge of the latest checkpoint, and rejects the abandon if there is a missmatch.

This can prevent a class of failures where pali had a stale view of the checkpoint and caused wiredtiger to step up from a non-latest checkpoint.