Add Evaluate overload that takes IDataView

[maryamariyan]

From @eerhardt on May 1, 2018, 9:52 AM PDT

See comment From @glebuk created May 1, 2018, 8:52 AM PDT for file src/Microsoft.ML/Models/BinaryClassificationEvaluator.cs on line below:

public BinaryClassificationMetrics Evaluate(PredictionModel model, ILearningPipelineLoader testData)

Evaluate [](start = 43, length = 8)

We should add another overload to evaluate against a scored IDV. There you'd need to specify columns for actual, predicted labels and score. (ok to ship after build) #Closed

[bartczernicki]

+1 this would be a big step up.

[shauheen]

With #371 this will not be needed anymore. Closing this issue.