Better data structure for the process tree

Currently, we rely on a plain list `ObservableCollection Processes` model.

This is problematic in some ways.

1. It's unclear how to efficiently find the process whenever we decide to delete it.
2. It is unclear how to produce the tree structure from them.

Currently, I am thinking about declaring some sort of observable dictionary (is there such a class in BCL? if not, we could create our own).

Make sure to review and eliminate every instance of `TODO[#5]` in the application code.