Sub-millisecond ID test failed once

[cuibonobo]

I thought I had addressed all of the bugs in the sub-millisecond ID generation, but it seems like there is still a rare/subtle one still in there. I didn't take a screenshot but what I observed when the failure happened is that the timeStampSet appeared to have two items (there should have been only 1) where the second item was the same as the first but with an appended zero (0).

I'm not sure how it overflowed... Might need to run the sub-millisecond tests a bajillion times to see what the actual error was.

Also, the test for detecting that an overflow happened is not well designed: expect(e instanceof Error).toBe(true) is not specific enough to determine which error was thrown.