Note a checkpoint is about to begin. Most beans will ignore,
but some can use this to wrap up tasks that shouldn't be
half-done during a checkpoint (and hold a lock to prevent
tasks from beginning during the checkpoint).
Do the actual checkpoint. Beans should ensure any state that
they would need to recover gets saved in an appropriate place.
A moderate amount of state may be saved as a JSONObject into
the Checkpoint (which then keeps it in the checkpoint directory.)
Larger amounts of state may be stored in a manner private to
the bean, or via other collaborating beans (such as BdbModule)
which checkpoint backgin database state.
Used to inform a bean that it should restore its state from
the given Checkpoint when launched (Lifecycle start()). May be
autowired or configured after build, but before launch, for
example via CheckpointService.setRecoveryCheckpointByName().