Currently, failures rolling back Changes which are managed by the BackupController may cause any subsequent restores to not be performed.  (The individual RestorableChange is able to decide whether an error will cause the rollback to stop but the default is an error in the RestorableChange will cause rollback to stop). 

We have decided to change that.so each error is treated as isolated.  A failure in one will not affect the others.  We will try to restore as much as possible.

We need to create a catchall exception handler in the BackupController to make sure an exception raised by any given Change will not cause the rest of the  Changes to not be run.

Acceptance Criteria:

• When backup_control.pop_all() is called, a failure in an early Change restore will log and error but the rest of the Changes recorded in backup_control will be performed.