Introduce a data model for "proctoring artifacts", aka the pictures
and audios, which so far have only lived as files in the filesystem
named after a convention
The new datamodel allows to quickly and reliably retrieve the pictures
collected by proctoring and also provides for a technical space, in
form of a JSON metadata column, to store additional information coming
from e.g. postprocessing phases.
The idea is to use this feature to provide reviewing tools of proctoring results and allow for flexile downstream postprocessing.
The file naming convention has not been changed for the time being, so this would be
retro-compatible. However, integrators should start relying on the new
data model to retrieve pictures.
