[CBM] Potential race condition in Rift Restore pipeline

What's the issue?
The crux of this issue is that our worker pool doesn't accept a context to handle clean teardown. In lieu of this, we have a context that we use to perform clean teardown.

// done - Handles the error returned by a worker, canceling the context if non-nil.

done := func(err error) error {

	if err == nil {

		return nil

	}

	cancel()

	return err

}

The error here, is that the context is canceled prior to the error being returned; depending on how goroutines are scheduled, the error returned by the worker pool may be:

1. This error
2. Or the error returned by the worker that tears down first (which is likely to be <nil>)

What's the fix?
We should probably have the worker pool properly use contexts to support clean cancellation.