Autograder fails and then passes with the same code

Hi! Upon pushing my code with the latest update of the autograder (the version from around 10pm), the test for flipping many coins with a random number argument failed, even though the output did not appear to deviate from what was required of the assignment:
::error::The output for test Flip many coins with random number argument did not match%0AExpected:%0A(\[|\{)\s?["']?(heads?|tails?)["']?\s?:\s?["']?\b[12345][0-9][0-9]\b["']?(,?\s?["']?(heads?|tails?)["']?\s?:\s?["']?\b[12345][0-9][0-9]\b["']?)?\s?(\]|\})
Confused on what my error was, I reran the autograding workflow and passed that test. Is there a reason (such as timing out) why the test might have failed and then passed with the same code?