Discuss deviance residuals in the class metrics chapter

Some references to dive into

• Pregibon, D. 1981. "Logistic Regression Diagnostics." The Annals of Statistics 9 (4): 705–24. (Section 4.2 defines the signed deviance residual d_i = ±sqrt(2 * [l(saturated) - l(fitted)]) used in the text.)
• McCullagh, P., and J. A. Nelder. 1989. Generalized Linear Models. 2nd ed. Chapman & Hall. (Section 2.4 covers unit deviances for general likelihoods, which is the basis for the multiclass extension.)
• Dunn, P. K., and G. K. Smyth. 1996. "Randomized Quantile Residuals." Journal of Computational and Graphical Statistics 5 (3): 236–44. (An alternative that stays signed and approximately normal for discrete (including multiclass) outcomes.)
• Liu, D., and H. Zhang. 2018. "Residuals and Diagnostics for Ordinal Regression Models: A Surrogate Approach." Journal of the American Statistical Association 113 (522): 845–54. (A signed alternative when the classes are ordered.)

Start with the class metrics chapter and review the four cited references, especially the sections on signed deviance, unit deviances, randomized quantile residuals, and ordinal models. Done means the chapter clearly discusses the relevant residual choices and their multiclass or ordinal context, with the references incorporated.