Filtering across multiple columns, best practices? Many-to-Many or 1-to-Many (dim)

I have a case for something where either many-to-many or a 1-to-many relationship seems like my only options and am seeking advice on best practices and which to use:

My example is user-manager assignments and specific filtering criteria:

I have a table of users that all report up to “Thomas”, some users may have multiple levels of managers in-between them.  My table looks like this:

I also have a fact table of claims, with a unique key, and assignment, among other details.

The ask: For a filter in the filter pane or slicer where if the user selects a name, they see their claim assignments, but also the assignments of everyone any level below them.

Example:  Thomas would see everyone, David sees David, Francis, Jamie, and Mary, Mary only see Mary.

I have only come up with 2 solutions, a many-to-many relation, a bi-directional relationship with the many side of the 1-to-many being the dimension table.

Note: I am aware of a third somewhat solution of using a measure to check for selections and a filter applied to each visual, however, that causes some visuals to time out, so I excluded that.

I have two Many-to-many ideas:

Concatenating and repeating all combinations in both tables.

This seems inefficient, as it adds a column with a lot of characters.

OR

Creating a table that repeats users for all combinations and joining on the analyst name:

1-to-many example:

Repeating claim number and user, and join on claim number:

A drawback to the 1-to-many, is that the dimension table could reach over 10M rows in some models I wish to use this filtering concept.

Natruatlly, avoid both of these things in a data model and use a traditional star schema; however, it seems unavoidable for this ask. 

What method would be considered the best practice?