Unlock: Chi-Squared Concentration

Two-sided exponential concentration for chi-squared sums of squared standard Gaussians: P(|Z/k - 1| > t) <= 2 exp(-t^2 k / 6) for t in (0, 1/2). Drives sub-exponential tails, variance-component inference, and Lipschitz Gaussian concentration.