save_control only works with single batches

[marcmaxson]

• need to update pipeline to handle multi-batch processing and saving controls for each batch
• loads manifest for each batch, which is slower, but allows run_pipeline to process separate batches as separate array types, which is more common issue with older GEO datasets. But upstream, need to create a step that looks at samples and puts them into separate batches if multiple array types are present. AT the moment, you have to manually split these datasets into different folders and run methylprep on each separate folder