If you retrieve an object in a microflow, you don't retrieve every object that it is associated with at the same time. This is due to basic memory and time constraints, you don't want to query the entire database when retrieving a simple object.
To be able to model something like this would introduce a lot of modelling complexity, you'd have to enter how 'deep' you want the references to be retrieved in a microflow in order to be able to show this information in the grid. It would also make your queries take longer, especially since it would have to retrieve the entire object chain and can not skip steps in associations unlike a datagrid with a database source which can optimize these things. Additionally it would have to retrieve full objects instead of just the attributes it needs.
So there are some reasons why this is not technically impossible but still has some serious implications. In addition, everything takes time to implement and some things are more wanted than others, another reason why features might not be available :)