MRG, ENH: Reduce memory usage of Welch PSD

[larsoner]

This:

# -*- coding: utf-8 -*-
import os
import mne
sample_data_folder = mne.datasets.sample.data_path()
sample_data_raw_file = os.path.join(sample_data_folder, 'MEG', 'sample',
                                    'sample_audvis_raw.fif')
raw = mne.io.read_raw_fif(sample_data_raw_file)
raw.plot_psd()
data_path = mne.datasets.brainstorm.bst_raw.data_path()
raw_path = (data_path + '/MEG/bst_raw/' +
            'subj001_somatosensory_20111109_01_AUX-f.ds')
raw = mne.io.read_raw_ctf(raw_path).load_data()
raw.plot_psd()

Goes from this on master:

To this:

@agramfort feel free to try it on mne-tools/mne-bids-pipeline#155

[hoechenberger]

@larsoner This does indeed fix our issue with the study template!

[larsoner]

I think you reintroduced a codestyle error I had fixed in master

[larsoner]

Ahh never mind, guess the "for ext in exts" is okay

[hoechenberger]

Thanks @larsoner!