Missing 'thresholds_and_metric_scores' API for binomial model metrics

[maurever]

We cannot call model.thresholds_and_metric_scores() or performance.thresholds_and_metric_scores() with Binomial models and metrics.

>>> from h2o.estimators.gbm import H2OGradientBoostingEstimator
>>> pros = h2o.import_file("https://raw.github.com/h2oai/h2o/master/smalldata/logreg/prostate.csv")
>>> pros["CAPSULE"] = pros["CAPSULE"].asfactor()
>>> train, test = pros.split_frame(ratios = [.75], seed = 1234)
>>> model = H2OGradientBoostingEstimator(distribution = "bernoulli",
...                                      ntrees = 100,
...                                      max_depth = 4,
...                                      learn_rate = 0.1,
...                                      seed = 1234)
>>> model.train(y = "CAPSULE",
...             x = ["AGE", "RACE", "PSA", "GLEASON"],
...             training_frame = train)
>>> model.thresholds_and_metric_scores()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/htillman/miniconda3/lib/python3.8/site-packages/h2o/utils/metaclass.py", line 352, in __getattr__
    return getattr(new_clz, name)
  File "/Users/htillman/miniconda3/lib/python3.8/site-packages/h2o/utils/metaclass.py", line 397, in __getattribute__
    return type.__getattribute__(cls, name)
AttributeError: type object 'ModelBase' has no attribute 'thresholds_and_metric_scores'

Thanks, @hannah-tillman, for discovering this bug.