Unstable IDs due to "hash of question text" algorithm

When the text of a question is updated, the question no longer has the same ID, breaking "permanent links" to it.

A stable way to refer to questions could also be useful for the proposed use case of generating exams from the .q files (For example, one could declare an exam as a list of questions. When the exam is recompiled from this representation, changes to the questions would be applied while the set of questions used would be the same.)

If the .q format is kept in the future, I propose adding some way to add a permanent identifier (like a UUID) to each question