That's an interesting approach.
Did your previous system also do instant grading of submissions, allowing students to immediately correct their code during the exam? My feeling is that giving students the option of resubmitting some time after the exam might not be as relevant when using CodeRunner. When students submit a CodeRunner answer that gets marked wrong, they normally try to fix it there and then. They're very reluctant (often too reluctant) to move on to the next question until they've got the marks for the one they're working on. I'm not sure how their behaviour would change if you were to say to them "You can come back after the exam closes and fix your program, paying a penalty that depends on how many changes you make". Would giving them that option really improve the quality of the final assessment?
Tim (below) has suggested a way in which you could give students the option of resubmitting after the exam has closed, but you would have to manually regrade the submission(s), applying your difference-dependent penalties. You could implement an off-line script to automate this using the exported responses both straight after the exam and again after the resubmission window. However, I'm not sure it would be worth the effort.
But it's certainly interesting to think about these things. Thanks.