That's just because Chevyshev bounds is a very weak general statement about all distributions.

High Energy Physics sigma is calibrated to match normal distribution quantiles.