This will let users to run their PyTorch code without network access,
provided that they have already cached the data.

### How has this been tested?
<!-- Please describe in detail how you tested your changes.
Include details of your testing environment, and the tests you ran to
see how your change affects other areas of the code, etc. -->
Unit tests.

Extracted from https://github.com/opencv/cvat/pull/5083

- Added a default arg for task data uploading
- Added an option to wait for the data processing in task data uploading
- Moved data splitting by requests for TUS closer to the point of use

This is useful for people whose home directory is too small/not fast
enough. It also lets us make the tests less hacky.

* Make the extractors return tensors instead of Python data structures.
* Let the user specify custom label IDs.