I think it is good that your final university project is encouraging you to contribute to an existing Open Source project, rather than to build something completely separate. When you work with Open Source software, it is essential to communicate with other developers, to work out the best way to do things, so hopefully the rules for your project allow that. (If not, the problem is with the rules, not with talking to to us ).
If you are going to talk with other developers about the project, I would have thought it is much safer to have the discussion in a public place like this. Then you can give the URL of this discussion in the references of your project, and your teachers can check (if they wish) that you have only asked for appropriate sorts of help, and that we have not done the project for you, we have only offered appropriate guidance. If you have any doubts about this, you could ask one of your professors to review this thread now.
--------
Now to answer your actual question, which is basically "When to load the data from the .CSV file and store it in the database?"
There are really two options, and I think they would both work.
1. When the teacher creates or edits the question, and clicks save, then load all the .CSV data for all variants and store it in the database. Then start_attempt can just load the data from the database.
2. In start_attempt, load all the .CSV data for the current variant and store it in the database then.
I think both options would work, so you can choose the one that is easier.