a) Simply run a test you know will fail back to back with and without your screen shot.
b) We had limited QC drive space. Only saved the screenshot of the last run on a network drive. Continually overwrote the same file. The script was either debugged before the next run or a defect against the application was logged with the screenshot. There was no need to maintain a big backlog in QC.
It may only turn out to be a second or so of time savings. But if you have tens of thousands of tests, that certainly adds up.