Fix eta-squared calculation in repeated measures ANOVA design

[raphaelvallat]

This PR fixes the issue described in #251

• In one-way ANOVA design, the calculation of the eta-squared n2 is now the same as the generalized eta-squared ng2. Previous versions of Pingouin used the same formula for the eta-squared and the partial eta-squared np2, which is invalid. Note that this invalid behavior is currently the default in JASP, but not in JAMOVI.

• The default effect size metric in repeated measures ANOVA design is now the generalized eta-squared ng2.

• The calculation of the eta-squared n2 in two-way repeated measures ANOVA design was invalid (approximatively similar to the partial eta-squared). This has been fixed.

• All effect sizes (n2, np2 and ng2) have now been tested against JAMOVI, for both one-way and two-way repeated measures design.

I am looking for one or two reviewers to review this PR.

Thank you,
Raphael

[codecov]

Codecov Report

Merging #254 (ce4b7d3) into master (8825a21) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head ce4b7d3 differs from pull request most recent head a36d3ed. Consider uploading reports for the commit a36d3ed to get more accurate results

@@            Coverage Diff             @@
##           master     #254      +/-   ##
==========================================
- Coverage   98.99%   98.99%   -0.01%     
==========================================
  Files          19       19              
  Lines        3287     3286       -1     
  Branches      523      523              
==========================================
- Hits         3254     3253       -1     
  Misses         17       17              
  Partials       16       16              

Impacted Files	Coverage Δ
pingouin/power.py	100.00% <ø> (ø)
pingouin/parametric.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8825a21...a36d3ed. Read the comment docs.

[raphaelvallat]

Still looking for 1-2 reviewers to review this PR