Multi sample test files for ATAC components

[VladimirShitov]

As pointed out by @bio-la , one-sample and multi-sample scenarios are quite different for ATAC. We should, therefore, have test files with multiple samples. Fabi suggested 2 options:

1. TEA-seq data: https://elifesciences.org/articles/63632
2. NeurIPS/Kaggle competition multiome data. I need to check with Christopher Lance where to get the raw data for that.
3. Another option would be to generate fake multi-sample data from the tiny bcl that we have. This is a bit toy example, but it would cover at least the easy scenario