feat: continuous recording (resilient against failing streams)

[ErikBjare]

This might be of interest to you @JohnGriffiths

Note: not actually 'continuous', it just saves every 10s. Ideally, it'd append rows into the CSV file instead of rewriting the entire file (but since a DataFrame is used this would require more extensive rewriting).

TODO

• Needs rebasing on upstream (remove unrelated commits)
• Needs testing
• Changes by black should be applied in separate commit, or skipped altogether
• Don't write the entire file every time

[sjawhar]

For appending, all you'd need to do is check if the file already exists and continuous mode is on. If so, use kwargs mode='a', header=False in the call to to_csv()

[xloem]

I've merged this into https://github.com/xloem/muse-lsl .

[ErikBjare]

Just finished this up with the additional comment from @sjawhar.

Works great now, ready for merge! (plz review @jdpigeon)

[jdpigeon]

Awesome. Thanks so much for this. Our original implementation was really quite shaky for collecting longer sessions.