Build the delay schedule for a retrying task. For attempt n (0-indexed) the base delay is base * 2**n (exponential backoff), but: - each delay is capped at cap, - you stop once adding the next delay would push the cumulative delay over budget, - and you never produce more than max_retries delays.
Return the list of delays actually scheduled.
What the tests check
3 on Run · 4 on Submit01exponential growth, no limits hit
02individual delay capped
03stops when budget would be exceeded
Submit runs these and one more check you don’t see, so code written for the visible tests alone can still miss.