S1/S2 spins NaN

From @mpgalleg comment on the Population Analysis Framework:

I tried to follow the tutorial and ran into an error when trying to calculate the observable population of BBH mergers using a population ran with NN. I did not run into this problem when using a population ran with IF.

ValueError: Input X contains NaN.
KNeighborsRegressor does not accept missing values encoded as NaN natively. For supervised learning, you might want to consider sklearn.ensemble.HistGradientBoostingClassifier and Regressor which accept missing values encoded as NaNs natively. Alternatively, it is possible to preprocess the data, for instance by using an imputer transformer in a pipeline or drop samples with missing values. See https://scikit-learn.org/stable/modules/impute.html You can find a list of all estimators that handle NaN values at the following page: https://scikit-learn.org/stable/modules/impute.html#estimators-that-handle-nan-values

Seems like it is breaking when trying to predict points between the grid of sensitivity? @ZepeiX also had a similar error, but I do not know if it was with NN or IF population.

update:
the problem is happening because some S1_spin and S2_spin are nan, which causes chi_eff values to be nan. Nan is provided to the function predict_pdet in selection_effects.py

It is happening to 142/41882 merging binary black holes. These BBH mergers are not specific to a particular formation channel or metallicity. They are coming from formation channels
‘ZAMS_oRLO1_CC1_oRLO2_CC2_END’, ‘ZAMS_oRLO1-contact_CC1_oRLO2_CC2_END’, ‘ZAMS_oRLO1_CC1_oRLO2_oCE2_CC2_END’, ‘ZAMS_CC1_oRLO2_oCE2_CC2_END’ with metallicities 0.1 , 0.01 , 0.001, 0.2.

Because the spins are NaN, this is not an issue with the population part, but with the output of the population run. This requires more triaging to figure out what's going on.

@mpgalleg, would you be able to provide an example binary that has this issue from a NN population? Feel free to provide a file with the population (on slack) that has this issue.