Can Transformer Attention Spread Give Insights Into Uncertainty of Detected and Tracked Objects?