Question: Is catastrophic cancellation an issue when calculating dot products of floating point vectors? And if so, how is it typically addressed?