Let $\bar{x}$ be the mean, $M$ be the median, and $\sigma$ be the population standard deviation of a dataset. According to the Hotelling-Solomons inequality, what is the tightest upper bound on the absolute difference between the mean and the median divided by the standard deviation?