Inconsistency in usage of LossFunction for AveragedPerceptron and OnlineGradientDescent

[abgoswam]

The Options class for AveragedPerceptron and OnlineGradientDescent have a field LossFunction which is used as a factory

machinelearning/src/Microsoft.ML.StandardLearners/Standard/Online/AveragedPerceptron.cs

Lines 42 to 46 in cabf55b

	public sealed class Arguments : AveragedLinearArguments
	{
	[Argument(ArgumentType.Multiple, HelpText = "Loss Function", ShortName = "loss", SortOrder = 50)]
	public ISupportClassificationLossFactory LossFunction = new HingeLoss.Arguments();

machinelearning/src/Microsoft.ML.StandardLearners/Standard/Online/OnlineGradientDescent.cs

Lines 38 to 43 in cabf55b

	public sealed class Arguments : AveragedLinearArguments
	{
	[Argument(ArgumentType.Multiple, HelpText = "Loss Function", ShortName = "loss", SortOrder = 50)]
	[TGUI(Label = "Loss Function")]
	public ISupportRegressionLossFactory LossFunction = new SquaredLossFactory();

Instead of using the factory pattern, they should be defined as IClassificationLoss and IRegressionLoss respectively.

That's what the public API also uses:

machinelearning/src/Microsoft.ML.StandardLearners/StandardLearnersCatalog.cs

Lines 167 to 172 in cabf55b

	public static AveragedPerceptronTrainer AveragedPerceptron(
	this BinaryClassificationContext.BinaryClassificationTrainers ctx,
	string labelColumn = DefaultColumnNames.Label,
	string featureColumn = DefaultColumnNames.Features,
	string weights = null,
	IClassificationLoss lossFunction = null,

Refer to test case OnlineLinearWorkout

@sfilipi