Use edfio for reading and writing edf files

[hofaflo]

edfio could replace PyEDFlib and mne for our use cases. We would have to drop support for Python 3.8, but that is in line with the schedule we decided on in #16 anyway.

[cbrnr]

This is great! Thank you for this PR, I've only found a typo!

[cbrnr]

Really? We need pooch just to get the SciPy ECG toy data?

[hofaflo]

Apparently yes 😬 Do you prefer to bundle an ecg signal with the package instead?

[cbrnr]

Do you prefer to bundle an ecg signal with the package instead?

Yes. I'd store it as compressed .npz, then it takes up only 163kB. We could store it in sleepecg/data (in analogy to sleepecg/classifiers)?

[cbrnr]

Maybe we could even add a little function to access the toy data (sleepecg.get_toy_ecg() or something similar)?

[cbrnr]

Regarding dropping support for Python 3.8, I don't think we can already remove from __future__ import annotations, right? This is still required for Python 3.9, or is it?

[cbrnr]

Of course the other option is to export the toy data to EDF and add edfio to our base requirements.

[hofaflo]

I opted for the .npz file because the .edf would be a bit larger and introduce slight signal deviations.

Ad future imports: A few have become obsolete, but most of them have to stay until 3.10, yes.

[cbrnr]

Great! I assume nothing has to be done to make sure the data gets added to the wheels, correct? We're still using MANIFEST.in unfortunately.

[hofaflo]

I think graft sleepecg makes sure this is included, yes. As cibuildwheel runs pytest after installing the built wheel, the checks would fail otherwise anyway.

[cbrnr]

Thanks @hofaflo!