Don't send extra data when getting the next delayed timestamp

[dcosson]

I had a problem where the scheduler stopped for a while so the delayed jobs started backing up. When the scheduler came back up each call to get_next_timestamp was returning the full set of timestamps, which became huge. I was burning through the backlog at a rate of only a couple of jobs per second; after applying this patch that jumped to a couple hundred.

[Whitespace]

:thumbsup:

[binarymatt]

thanks