Still don't quite understand rationalle for keeping data in layers before integration

[wudustan]

I'm still getting used to v5 but I don't understand the reason for why data from different layers is kept saparate through integration. Can someone explain why the dataset isn't just flattened at merge like with v4 and what the benefit is.

For example, I have data from my own 10x scRNA, data from a paper and data from a public portal. Rather than integrating across different technologies, I am seeking to integrate across different sources of data and handle the technical aspects of those batches.