ENH: set color for bad channel with spatial_colors=True

[larsoner]

Closes #12137

@hoechenberger can you try it with your data to see if it helps?

This will affect at least evoked.plot and spectrum.plot as spatial_colors=True by default.

• See if it actually helps with reports
• Add test
• Add changelog entry

[hoechenberger]

@larsoner i don't know when I'll will have time to test this, don't wait for me 🥹

[larsoner]

On main:

import mne
raw = mne.io.read_raw_fif(mne.datasets.sample.data_path() / "MEG" / "sample" / "sample_audvis_filt-0-40_raw.fif")
raw.pick("eeg", exclude=()).load_data()
raw.plot_psd(exclude=())

gives:

On this PR it's:

There is a bug on main where bad_ch_idx=[] so the dashed-ness was never applied. But I like that in this PR that is applied and it's gray -- I think it works well.

With the gray we could consider getting rid of the dashing which makes the bad one easier to resolve:

[hoechenberger]

I like the last figure you posted!

[larsoner]

Ready for review/merge from my end.

When I modified the Report tutorial with dashing turned off it looked like this. So I decided in the end to keep it to make it slightly more differentiable.

[hoechenberger]

👌