fix leaf_ids type error

[haribaskar]

fixed TypeError: object of type 'numpy.int32' has no len()

File "C:\Users\username\test.py", line 20, in churn_factor_xgb
    prediction_text = explain_prediction(model,X_test.iloc[i])
  File "C:\ProgramData\Anaconda3\lib\site-packages\singledispatch.py", line 210, in wrapper
    return dispatch(args[0].__class__)(*args, **kw)
  File "C:\ProgramData\Anaconda3\lib\site-packages\eli5\xgboost.py", line 196, in explain_prediction_xgboost
    booster, dmatrix, n_targets, feature_names, xgb_feature_names)
  File "C:\ProgramData\Anaconda3\lib\site-packages\eli5\xgboost.py", line 247, in _prediction_feature_weights
    assert len(tree_dumps) == len(leaf_ids)
TypeError: object of type 'numpy.int32' has no len()

[lopuhin]

Hi, thanks for the pull request, but it looks like the fix is not quite correct, xgboost tests are failing. It would be great if you could also add a test for this issue, or tell how to reproduce it.