Bug fix and small improvement

[matsvanes]

Fixes ica_label, and added an mne wrapper to preprocessing module for interpolating bad channels (with reset_bads=False by default - needed to keep track of bad channels)

[matsvanes]

Actually, it might be better to find another workaround for interpolate_bads.. Because it reset_bads=False, it might affect functionality going forward (i.e., if only good channels are selected). What do you think @cgohil8 ?

[cgohil8]

How come we have to set this to False now when before it defaulted to True. Was there always a bug or did something else change so we need to pass reset_bads=False? I think the main questions are:

• How does it affect the ICA? We don't want bad channels to be included.
• How does it affect the covariance calculation in the source reconstruction? I actually can't recall how bad channels are dealt with, i think it just includes everything.