I'm not sure if this an improvement in terms of structure and readability. Some (very small) parts of the drag and resize logic are now "hidden" in the manager class, while the majority of it stays in the mixins. The class itself doesn't encapsulate any logic on its own, but just contains some utilities and some state. Its purpose is not clear from the name and it's also hard to come up with a better one. The weak map was also confusing me at first.

I think I prefer the duplication in this case, makes it easier to understand what's going on. As an alternative, a random idea would be to merge the two mixins. Or extract some utility functions that avoid duplicating some of the calculations at least. But I don't see a lot of value here in general, unless I'm missing some context where we would need to introduce more duplication for a new feature.

Thanks, actually I was also thinking about merging mixins. It's probably ok to keep the code the way it is.

I still think we should revert some changes from the base styles PR and update tests accordingly.