How should I split the entity data if calculate dfs max_depth more than 1 on spark?

### Feature Request Title

I refer to [this article](https://community.alteryx.com/t5/Data-Science/Featuretools-on-Spark/ba-p/546383) to use spark on my project, but I have a problem about if calculate dfs max_depth more than 1, how to split the data ?

My data model just like the pic below. And I want to calculate Entity_0's features, with max_depth = 2. If I just split all the entity with Entity_0's id, the relation relation_3_1 and realtion_4_2, will it be wrong?


Should I filter Entity_3's id in Entity_1 and Entity_4's id in Entity_2, and put the data together?

Thanks a lot