ENH: Preprocessed raw data

[larsoner]

Ideally MNE-BIDS-Pipeline would be usable for all possible analyses. However, currently (and for the foreseeable future) it's used for event-based designs. For things like naturalistic speech or resting state analyses (or other types of analyses probably, too) it would be useful to have not just preprocessed epochs but also preprocessed raw data available for downstream analyses not yet implemented. So assuming we want to add it, we have two options I think:

1. Always preprocess the raw data the same way we do for epochs, and save the resulting files
2. Make saving preprocessed raw data optional / opt-in

I'm inclined toward (1) but it could be wasteful for some people in terms of disk usage. Thoughts?

[larsoner]

... in other words, after creating sub-01_task-whatever_run-01_proc-filt_raw.fif create something like a sub-01_task-whatever_run-01_proc-clean_raw.fif or similar