Used for implementing custom policies in case of communication failures with a remote site. The handle methods are
allowed to throw instances of BackupFailureException to signal that they want the intra-site operation to
fail as well. If handle methods don't throw any exception then the operation will succeed in the local cluster. For
convenience, there is a support implementation of this class: AbstractCustomFailurePolicy
Lifecycle: the same instance is invoked during the lifecycle of a cache so it is allowed to hold state between
Threadsafety: instances of this class might be invoked from different threads and they should be synchronized.