To remove the rows that you annotated with "Duplicates", you could use the Views Distinct module. From its project page:

Relationships or other joins in Views often create "duplicate" results. For example, a node with a field that has multiple values may show up in the View once per value in the multi-value field. It's frustrating, and the "DISTINCT" SQL query option in the Views UI does not actually solve the problem because the result row is technically distinct. This module aims to give a simple GUI method to remove or aggregate these "duplicate" rows.

I can't think of any other missing piece to complete your puzzle.

To remove the rows that you annotated with "Duplicates", you could use the Views Distinct module. From its project page:

Relationships or other joins in Views often create "duplicate" results. For example, a node with a field that has multiple values may show up in the View once per value in the multi-value field. It's frustrating, and the "DISTINCT" SQL query option in the Views UI does not actually solve the problem because the result row is technically distinct. This module aims to give a simple GUI method to remove or aggregate these "duplicate" rows.

Assuming that would indeed remove the duplicate rows, the "What I get" should match with "What I try to achieve". So with that the setup of your relationships like you already defined them should be OK also (I can't think of any other missing piece to complete your puzzle).

You could use the Views Distinct module. From its project page:

Relationships or other joins in Views often create "duplicate" results. For example, a node with a field that has multiple values may show up in the View once per value in the multi-value field. It's frustrating, and the "DISTINCT" SQL query option in the Views UI does not actually solve the problem because the result row is technically distinct. This module aims to give a simple GUI method to remove or aggregate these "duplicate" rows.

To remove the rows that you annotated with "Duplicates", you could use the Views Distinct module. From its project page:

Relationships or other joins in Views often create "duplicate" results. For example, a node with a field that has multiple values may show up in the View once per value in the multi-value field. It's frustrating, and the "DISTINCT" SQL query option in the Views UI does not actually solve the problem because the result row is technically distinct. This module aims to give a simple GUI method to remove or aggregate these "duplicate" rows.

I can't think of any other missing piece to complete your puzzle.