Allow Task.par(Iterable) to fail-fast

Is there any way of failing fast when executing a list of tasks in parallel?

For my use-case I want to cancel all the remaining tasks when one of the tasks in the parallel group fails.
I saw that `Task.par(Task, Task, ...)` behaves like this.
I wonder if `Task.par(Iterable)` can support it as well?

Start by comparing the existing Task.par(Task, Task, ...) behavior with Task.par(Iterable), focusing on how a failure cancels the remaining tasks. Done means the iterable form fails fast and cancels outstanding tasks when one task fails, with coverage for that behavior.

Written by the indexing model from the issue text.