Fix file naming in outlier_detection

[jdavies-st]

Currently in outlier_detection, the writing out of intermediate files is handled by subroutines or subroutines of subroutines. Ideally, this should happen in the OutlierDetectionStep.process method at the top level and be done using the standard Step.save_model methods.

This will require a significant refactor of where most of the logic of the step happens. Currently it happens in the OutlierDetection class, but it really should be happening in the OutlierDetectionStep class.