Exercise `ucurve_fit` strange behavior for `absolute_sigma=True` if `sigma=None`

[ChAr-De]

See this scipy issue. If we use the solution of the exercise without errors of the y parameter, the errors of the fit parameters are wrongly calculated. Maybe we should change the solution to also set absolute_sigma=False or add a comment to prevent the usage of the function if the input data has no errors.

[aknierim]

Just for clarification, you're talking about this line, right? https://github.com/pep-dortmund/toolbox-workshop/blob/52ed55a114bd64ee7b54bb734726023f388ba0f6/exercises-latex/13-python/curve_fit.py#L23

[aknierim]

I would argue for something like this then (ll. 13-16):

 if np.any(unp.std_devs(y) == 0):
        sigma = None
        abs_sigma = False
    else:
        sigma = unp.std_devs(y)
        abs_sigma = True

And then changing https://github.com/pep-dortmund/toolbox-workshop/blob/52ed55a114bd64ee7b54bb734726023f388ba0f6/exercises-latex/13-python/curve_fit.py#L23 to

absolute_sigma=abs_sigma,