DOC use TLCenter for example on RPA

[qbarthelemy]

Use TLCenter in example on cross-subject transfer with Riemannian alignment
https://moabb.neurotechx.com/docs/auto_examples/how_to_benchmark/plot_cross_subject_transfer_rpa.html#why-align-covariance-matrices

[bruAristimunha]

thanks again for this @qbarthelemy 🙏🏽

[bruAristimunha]

FYI: some simplification in the riemannian alignment @toncho11

[toncho11]

I am not sure, but I think there may be a semantic issue in fit_transform().

The previous code centered each source subject independently, while the new code calls TLCenter.transform(), which uses only the target-domain center. TLCenter explicitly documents that fit_transform() and fit().transform() have different semantics: https://pyriemann.readthedocs.io/en/latest/generated/pyriemann.transfer.TLCenter.html

I think this must be investigated before merging.

[qbarthelemy]

@toncho11 is right.
I fixed fit_transform so that it correctly re-centers each domain.

[toncho11]

Looks OK. Did you check if the results are similar between the old and the new version?

[qbarthelemy]

Yes, I had checked.