A common case I find myself dealing with is performing joins between two data sets just to filter out values of my primary set that are not in some reference list. In this case, I'm not interested in the attributes of that reference list, only that the joining process removes the entries that don't match.
In SQL, this would be the equivalent of something like this, with an inner join:
SELECT person.*
FROM person
INNER JOIN person_car ON person.person_id = person_car.person_id
...or in the more explicit way, with an actual semi join:
SELECT * FROM person
WHERE EXISTS (
SELECT * FROM person_car WHERE person.person_id = person_car.person_id
)
The benefit of this is that the attribute list of downstream features cannot be altered by the join, which is useful for dynamic workbenches and dynamic transformers.
This would be a simple matter of adding a "Discard supplier attributes" mode to FeatureJoiner and FeatureMerger.