If Quirrell has a time-turner it seems plausible the time cost required for finding SHA-1 collisions would be irrelevant and the entire method would be insecure.
Harry's work suggest that time-turners respond poorly to being used for computational short-cuts.
I would like to make a prediction, but I'm not ready to raise the subject of the prediction just yet. So to show that it was not just obvious in hindsight, and to strengthen my precommitment to write an article about it later, here is the sha1sum of my prediction: 9805a0c7bf3690db25e5753e128085c4191f0114.