FeaturizeText should allow only outputColumnName to be defined

[artidoro]

One of the extension methods for FeaturizeText needs both the outputColumnName and the inputColumnNames to be provided.

There is no compile error if inputColumnNames is not provided, only a runtime error.

We should fix the error so that when inputColumnNames is not provided, it is set to new[] { outputColumnName } as we do everywhere else in the code base.

machinelearning/src/Microsoft.ML.Transforms/Text/TextCatalog.cs

Lines 60 to 65 in c3bdaaa

	public static TextFeaturizingEstimator FeaturizeText(this TransformsCatalog.TextTransforms catalog,
	string outputColumnName,
	TextFeaturizingEstimator.Options options,
	params string[] inputColumnNames)
	=> new TextFeaturizingEstimator(Contracts.CheckRef(catalog, nameof(catalog)).GetEnvironment(),
	outputColumnName, inputColumnNames, options);

[SnakyBeaky]

The only similar case I've seen, is when inputColumnName is a single string and not a string[]. In those cases, the null-coalescing operator is used.

For example, in Microsoft.ML.Transforms/Text/TokenizingByCharacters.cs#L595 :

new[] { (outputColumnName, inputColumnName ?? outputColumnName)

But I have not seen this done when the input column names is an array and not a single string.