When I use the 'list_questions' method to access the questions of a survey via API, I have difficulties understanding how question types for subquestions are determined. We mostly use array type questions, which per here have the type 'F', but the subquestions seem to have random question types - sometimes 'F', sometimes 'T' ('T' is certainly completely inaccurate in this case). Can I somehow influence this with some settings in Limesurvey?