AnyClassifier fails on baseball

```python
from dabl.models import AnyClassifier
from sklearn.model_selection import cross_val_score
from sklearn.datasets import fetch_openml

bunch = fetch_openml(data_id=185)
X = bunch.data
y = bunch.target
cross_val_score(AnyClassifier(), X, y, scoring='f1_macro', cv=10, error_score='raise')
```
errors with a NaN error and I don't know why :-/
@thomasjpfan @NicolasHug feel free to check it out if you want to play with dabl.

Reproduce the failure using the provided fetch_openml(data_id=185), AnyClassifier, and cross_val_score example. Start by tracing AnyClassifier on the baseball dataset and inspect how the f1_macro scoring error becomes NaN; done means the same cross-validation call completes without the reported error and produces valid scores.

Written by the indexing model from the issue text.