EPI masking: minimize run-to-run difference using a soft mask

[oesteban]

Add a soft mask to MNI152NLin2009cAsym to replace the hard mask here:

niworkflows/niworkflows/func/util.py

Lines 291 to 292 in 76f831f

	brain_mask = get_template(
	'MNI152NLin2009cAsym', resolution=2, desc='brain', suffix='mask')

Having a probabilistic mask that can be binarized after resampling will certainly reduce the differences making our reference run-to-run unreliable.

[oesteban]

templateflow/tpl-MNI152NLin2009cAsym@d6753e2