We are currently (de-)serialising (`m_to_dict`, `m_from_dict`) repeating sub-sections as lists. As lists are basically "integer keyed dicts", we could offer dicts as an alternative. This could be done transparently for deserialising and per flag on serialising. Further, we have the concept of a "label_quantity" already. This is implemented very softly as a `more` quantity on sections. This is used by the archive browser in the UI to show values instead of numbers when presenting subsections. Further, we added specialised deserialisation for schemas to support dicts for the `quantities` and `sub_sections` sub section of `Section` and `section_definitions` and `category_definitions` sub section of `Package`. Maybe there is more for attributes? This is how this could get implemented more properly: - `SubSection` gets a new `key_quantity: Quantity` property that refers to a quantity in the SubSection's `sub_section:Section`. If the `sub_section` has a quantity with `more.label_quantity`, we use this as `key_quantity` for backwards compatibility. - When a sub section with `key_quantity` is set or deserialised, we check if the key quantity has unique values for all sub sections. - `m_from_dict` deals both with dicts or lists. - `m_to_dict` has three modes: always dict, always list, dict if `key_quantity` is set (default). - The UI deals with both lists and dicts. We remove the treatment of the old `more.label_quantity`. - We remove the specialised `m_from_dict` implementations on the schema classes `Section`, `Package`. This could also be beneficial for the new graph API #1180 which uses dicts to realise a sub-section-like relationship between uploads->entries, users->uploads, etc.