Does Looker no longer produce the `non-uniqe primary key` error when a pk is not unique?

[duke_of_data]

I wanted to demonstrate how symmetric aggregates work, and how, if the designated primary keys are not, in fact, unique, the algorithm returns a very large number and Looker flags it with a `non-unique primary key error`.  So I intentionally created a table with a duplicate key (I took the users table in the canonical thelook model and unioned in an extra user with an id that exists).

However, to my surprise, Looker just gave me the very large number as a result.  This seems …. less than ideal.