In plain gridsearch, why is this code here?

[neukym]

    n_samples_per_split = int(seconds_per_split * emeg_sample_rate * 2 // downsample_rate)

    # the number of samples in the transform 'trial' which is half that needed for the EMEG
    n_trans_samples_per_split = n_samples_per_split // 2

Why is there a multiply-by-2 and divide-by-2 here? What does it buy us? @neukym to ask @ollieparish.

[neukym]

@caiw has pointed out that this is because the duration of the EMEG split needs to be twice a long to do the pass over the function. 👍 Will confirm with @ollieparish.