There are lots of rather simple tests (e.g. Shapiro–Wilk test) checking for normality and I think it will be useful to have some of them in ClickHouse.
e.g. Pearson's correlations does not assume normality but outliers can have great influence on correlation coefficient.