[FEATURE] Allow subject argument to olink_one_non_parametric_posthoc when using Friedman's test

[sulealtintas]

Is your feature request related to a problem? Please describe.
It is unclear whether samples from the same subject are paired correctly in the post-hoc test for Friedman's test, as olink_one_non_parametric_posthoc does not take a subject argument. I found an old issue #161 on this, but it seems to have been marked completed without resolution.

Describe the solution you'd like
It should be possible to provide an argument subject specifying the column with the subject IDs to olink_one_non_parametric_posthoc, similarly to when calling olink_one_non_parametric. The data should then be sorted according to subject before the call to rstatix::wilcox_test with paired = TRUE, which assumes ordered data when pairing the data points.

Additional context
Example illustrating the importance of sample ordering in paired rstatix::wilcoxon_test: https://stackoverflow.com/a/78213490