_assert_equal_properties(data_input) about pyerrors

[fjosw]

Yes, I implemented this additional check to be sure that the data is not corrupted. I believe all entries of a correlator should be defined on the same set of configurations in all practical scenarios (although that is not strictly necessary for the correct error propagation). Do you have any use cases in which you would want to initialize such a correlator?

Also the json output for correlators strictly assumes that all non None entries are defined on the same set of configurations. Otherwise the reassembling into a 3D array would not work. That is also why I had to implement the function _nan_Obs_like (https://github.com/fjosw/pyerrors/blob/develop/pyerrors/input/json.py#L164).

[JanNeuendorf]

I could imagine someone measuring more configurations for one smearing level and still wanting to use the GEVP methods. Theoretically the new Obs can know, what configurations it was measured on (?) and be compatible. But i would not need this and if it is hard to implement, we can leave it as it is.

[fjosw]

I propose to leave the Corr.__init__ method as it is for now. If such a special case is needed in the future we can revisit this issue.