Customizing `processing_pipeline` for benchmarking

[Aceticia]

Hi all. I'm trying to benchmark a few pretrained EEG models, but most of these models have their own required processing pipelines and thus the automatically added band pass would usually need to be removed. I'm currently just using evaluators individually and overriding the processing pipelines according to this guide, but I think giving the user some basic choices for modifying the pipeline used in benchmarking would be beneficial.

[bruAristimunha]

I agree with you!

At the moment, we currently doing pipeline surgery, but this current solution is not very intuitive. You need to understand very well pipeline, moabb, and how we do the processing.

To be more community driven, can you please think how a more intuitive solution could be make in our opinion?

I feel util functions to manipulate this could be useful.

[bruAristimunha]

I push something within the artefact pipeline, but I get that the interface is not optimal... I will push something in this direction.