Call to .split_data() in .label() causing confusion

[AlainPilon]

The call to split_data is causing some confusion IMHO.

It should either:

• be a flag we can toggle on/off.
• copy the files to the YOLO train/val folders instead of moving them.
• have some output in the command line to tell the user about the file being moved.

Right now, someone calling label() would see the YOLO folder structure appears with an empty annotations folder and waste a few minutes trying to figure out why.

To make it even more confusing, calling label() a second time would crash on the image copy (because it does not want to overwrite existing file) and keep all the annotations in the annotations folder.

[capjamesg]

@AlainPilon Thank you for opening the Issue! I think label() should save annotations unless no objects of interest were found in all images. Can you confirm that Autodistill was able to label at least one image in your dataset? Also, can you share the folder path for the empty annotations folder?