Some scikit-learn estimators no longer work with array_function enabled

There's nothing dask-ml can do I think. Just filing this here till I can figure out the right thing to do.

```python
In [1]: import dask.array as da

In [2]: import dask_ml.datasets

In [3]: X, y = dask_ml.datasets.make_classification(chunks=50)

In [4]: import sklearn.linear_model

In [5]: clf = sklearn.linear_model.LogisticRegression()

In [6]: clf.fit(X, y)
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
in
----> 1 clf.fit(X, y)

~/Envs/dask-dev/lib/python3.7/site-packages/sklearn/linear_model/logistic.py in fit(self, X, y, sample_weight)
1536
1537 multi_class = _check_multi_class(self.multi_class, solver,
-> 1538 len(self.classes_))
1539
1540 if solver == 'liblinear':

TypeError: 'float' object cannot be interpreted as an integer
```

Previously, that would have been converted to an ndarray by scikit-learn and fitted.

Reproduce the failure using the Python session shown in the issue, starting with dask_ml.datasets.make_classification and sklearn.linear_model.LogisticRegression.fit. Inspect the array_function compatibility boundary to determine whether dask-ml or scikit-learn owns the behavior; the issue does not yet define an agreed fix or test for done.

Written by the indexing model from the issue text.