In the current implementation, the Retry instances only check if the next delay (sleep period) happens before the deadline. If the next sleep period is long compared to the remaining time, the last retry can happen "long" after the specified deadline, which can be undesired in certain scenarios.
As an API Core user I would like to have an option to mark the deadline as strict, making the Retry object to complete retrying (approximately) at the deadline, not beyond.