My main concern is the lack of gradual difficulty in the tutorials : reproducing research results is great, but gently steering the user might be missing.
While example 5 is simple as claimed, example 2 is rather complex. To engage scientists to use the software, step-by-step tutorial, or a better gradation of difficulty in the notebook examples, seem required to smoothen the steep learning curve.
Remark from https://github.com/openjournals/joss-reviews/issues/2973#issuecomment-771524127