Added catch for predictors that do not support summary()

[mstfbl]

Fixes #110 . Added a catch for the .summary() function in Pipeline.py for the FactorizationMachineBinaryClassifier.

[mstfbl]

On the ML.NET end, if a summary model is not obtained from any trainers, an IDataView is returned in the following way (in line 89, Microsoft.ML.Data\EntryPoints\SummarizePredictor.cs):

        if (dvGetter == null && rowGetter == null)
        {
            var bldr = new ArrayDataViewBuilder(env);
            var summaryModel = predictor as ICanSaveSummary;

            // Save a data view containing one row and one column with the model summary.
            if (summaryModel != null)
            {
                var sb = new StringBuilder();
                using (StringWriter sw = new StringWriter(sb))
                    summaryModel.SaveSummary(sw, schema);
                bldr.AddColumn("Summary", sb.ToString());
            }
            else
                **bldr.AddColumn("PredictorName", predictor.GetType().ToString());**
            summary = bldr.GetDataView();
        }

In the Python end, we receive a Pandas DataFrame with 1 column named PredictorName with the name of the given Predictor. Thus, unless the way of checking for the summary of a given model changes in ML.NET, we can check for the usability of the .summary() function in NimbusML by checking to see if the produced output equals this DataFrame.

[ganik]

PLs add test to show what FactorizationMachine error message is now

[mstfbl]

@ganik I am now specifically checking for the error raised by the FactorizationMachine classifier in the following way:

NimbusML/src/python/nimbusml/tests/model_summary/test_model_summary.py

Lines 119 to 124 in 208fe88

	try:
	pipeline.summary()
	except TypeError as e:
	self.assertEqual(e.args[0], "One or more transformers in this pipeline do not support the .summary() function.")
	else:
	assert False

[pieths]

See one minor comment about wording in exception message.