The problem is that the subquestion inside the

{1...} could include the half of a HTML code like <pre> or </pre>.
We could have an answer or a feedback that with that half part.
We need to HTML clean
the question text of the main question which contains the sub substitued as {#1},
the subquestion text which contains the {1...}

and the individual answers and feedback related to the

{1...}

.

When editing the question this could be done in the validation process and returning to user the clean code
with a message that his code have some HTML cleaning and he should verify if the result is OK.
If the user have done a copy and paste that introduces an incomplete HTML the editor will do the cleaning without any warning.

On import we have to decide if it a real error or just a silent modification.