Which mechanism is preferred for implementing timeouts in Swift structured concurrency?

Implementing Task Timeouts

Swift's structured concurrency model provides cooperative cancellation via Task.isCancelled and Task.checkCancellation(), allowing child tasks to halt execution when a parent is cancelled. However, the standard library does not provide a native timeout parameter for individual Task initializations.

To enforce a maximum execution time, developers often implement a race condition using a TaskGroup where one child task performs the actual work and another calls Task.sleep(nanoseconds:) to act as a timer. This requires manual management of the group to ensure the winning task cancels the losing one to prevent resource leaks.

Given the cooperative nature of this model, what is the most reliable pattern for ensuring a timeout triggers immediate cancellation of a long-running asynchronous operation? Does the use of a TaskGroup race introduce significant overhead compared to custom timer implementations?