Nan in the load_expanded_ensemble_case_1 dataset

[xiki-tempula]

One would assume that there shouldn't be any NaNs in the test data set, but simply removing the NaNs from the load_expanded_ensemble_case_1 dataset would make some tests fail. See #184

[orbeckst]

Do you suspect that there's something wrong with the data?

[xiki-tempula]

@orbeckst I think there are NaNs in the dataset as dropna would alter the size of the dataset.

[orbeckst]

Did you check that this is what happens? Then the next stage would be to contact the original authors to verify that these nans belong there. Once this is all confirmed we can consider changing the test answers.

[orbeckst]

@xiki-tempula please remind me, what do we have to do here? In PR #184 we learned that the NAN should not really be in the dataset. But with #171 solved (PR #183) we can use the dataset.

Maybe we only need to add a note in alchemtest that the dataset contains NAN?

[orbeckst]

Nothing else to do here...