Improve CI coverage for non-parametric CLES and rank differences

[mattansb]

Open a new issue / pr to discuss improving the CI covered for all of these methods - since they are all monotonic transformations of one another, it makes sense to me they can be improved together.

Originally posted by @mattansb in #496 (comment)

Also related to #479

[mattansb]

@arcaldwell49 feel free to contribute here if you'd like, there is not rush or pressure (:

[bwiernik]

The fisher z transformation applies to Pearson r and approximately to phi — it will undercover for any rank correlation or continuity correction like biserial, tetrachoric, or polychoric unless the n is adjusted downward

metafor::escalc has sampling error formulas for some of these I believe

[arcaldwell49]

Very good points @bwiernik . Also, I noticed an error in my code where I ran some simulations earlier this month.... The issue may be mute, but let me re-run the simulations to check.