Handle single-sample comparsion in pairwise_test

[George3d6]

Due to a scipy interface change the pairwise_test can fail when obtaining the t-value from a single sample comparison, this, I believe, fixes that issue.

[raphaelvallat]

Hi @George3d6,

Thank you for this!

1. Can you be more specific about which versions of SciPy have this potential bug? Did it appear in SciPy 1.9?
2. Does this issue only occur for one-sample T-tests?
3. There seems to be an unrelated unit test failure, I'll fix that now

[codecov]

Codecov Report

Merging #299 (9fe378e) into master (dce908b) will decrease coverage by 0.04%.
The diff coverage is 94.59%.

@@            Coverage Diff             @@
##           master     #299      +/-   ##
==========================================
- Coverage   98.75%   98.71%   -0.05%     
==========================================
  Files          19       19              
  Lines        3298     3335      +37     
  Branches      529      537       +8     
==========================================
+ Hits         3257     3292      +35     
- Misses         24       25       +1     
- Partials       17       18       +1     

Impacted Files	Coverage Δ
pingouin/multivariate.py	100.00% <ø> (ø)
pingouin/power.py	100.00% <ø> (ø)
pingouin/parametric.py	99.56% <33.33%> (-0.44%)	⬇️
pingouin/pairwise.py	99.46% <100.00%> (+0.05%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[George3d6]

I'm actually unsure when the scipy interface might have changed, or if this was always the case, I'll look into it.
But as can bee seen, the current scipy interface is returning a tuple that the code treats as expecting to be a float or int, so the impl is most certainly wrong

[George3d6]

@raphaelvallat as far as I can see this isn't a recent change in the scipy interface as I suspected, so I guess the bug was always there.

[raphaelvallat]

Got it, well I think we're good to merge. Thanks again for the PR!