MRG, BUG: Fix STC limit bug

[larsoner]

Close #7911

On master, auto control points for the vec STC are [1.86911522 2.26944519 6.37046324] and standard STC are [ 3.95048065 4.56941314 17.72451438]. On this PR, there are two fixes:

1. vector_alpha actually has an effect
2. _process_clim uses the .magnitude() data for vector mode, which makes more sense.

On this PR, this code:

import os.path as op
import mne

data_path = mne.datasets.sample.data_path()
sample_dir = op.join(data_path, 'MEG', 'sample')
subjects_dir = op.join(data_path, 'subjects')
inv = mne.minimum_norm.read_inverse_operator(op.join(
    sample_dir, 'sample_audvis-meg-oct-6-meg-inv.fif'))
evoked = mne.read_evokeds(op.join(sample_dir, 'sample_audvis-ave.fif'))[0]
evoked.apply_baseline((None, 0))
stc = mne.minimum_norm.apply_inverse(
    evoked, inv, method='dSPM', verbose='debug', pick_ori='vector')
initial_time = 0.1
# stc, kwargs = stc, dict(overlay_alpha=0.5, brain_alpha=0.5, vector_alpha=0.)
stc, kwargs = stc.magnitude(), dict(alpha=0.5, surface='white')
brain = stc.plot(subjects_dir=subjects_dir, initial_time=initial_time,
                 clim='auto', hemi='lh', views='lat',
                 smoothing_steps='nearest', verbose=True,
                 time_viewer=False, size=(600, 600), background='k',
                 **kwargs)
brain.show_view('lat')

• Run as is (magnitude mode):
  • Mayavi
    
  • PyVista
    
• Run in vector mode (second stc, kwargs line commented out) produces:
  • Mayavi
    
  • PyVista
    

The slight coloration of the brain itself differs because mayavi has backface culling on for the surface and our _Brain doesn't. I actually think our behavior is probably better for translucent brains.

There is a bug/mismatch where the limits are not computed properly in auto mode, probably related to whether or not the entire stc is used, or the initial time.

[larsoner]

@GuillaumeFavelier feel free to merge if you're happy, failure is just the 3.8 segfault bug. Also feel free to look into that one if you want, not sure why it started happening.

[GuillaumeFavelier]

LGTM!

[GuillaumeFavelier]

Thanks @larsoner