WIP: use get_feature_names_out from new sklearn if possible

[agramfort]

follow up on https://mne.discourse.group/t/future-warning-in-feature-extraction/4631/1

just to start the ball rolling here...

[codecov]

Codecov Report

Merging #77 (7abf7b4) into master (399092a) will decrease coverage by 0.05%.
The diff coverage is 80.00%.

@@            Coverage Diff             @@
##           master      #77      +/-   ##
==========================================
- Coverage   93.53%   93.48%   -0.06%     
==========================================
  Files          10       10              
  Lines        1455     1459       +4     
==========================================
+ Hits         1361     1364       +3     
- Misses         94       95       +1     

Impacted Files	Coverage Δ
mne_features/feature_extraction.py	94.73% <80.00%> (-0.54%)	⬇️

---

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 399092a...7abf7b4. Read the comment docs.

[etiennedemontalivet]

Maybe changing the 4 lines with just:
feature_names = extractor.get_feature_names_out()
will increase code coverage ? (if this is the issue why not merging the PR)

[etiennedemontalivet]

Is there any plan to merge this pull request ?
Since sklearn 1.2.1, they removed get_feature_names from FeatureUnion, so now it raises an error when using extract_features with return_as_df=True. Please tell me if I can help