On the bertron "project" (which uses a LinkML schema), we have added functions to the bertron-schema repo which expose the sample data to the outside world (i.e. to applications that import the "bertron schema" package). The tests targeting those functions are implemented here. The functions, themselves, are implemented here.

I think exposing sample data would facilitate testing downstream applications against schema-compliant and schema non-compliant data.

CC: @sujaypatil96, @turbomam