Generate built-in SweepableEstimator classes for all available estimators

[LittleLittleCloud]

We are excited to review your PR.

So we can do the best job, please check:

• There's a descriptive title that will make sense to other developers some time from now.
• There's associated issues. All PR's should have issue(s) associated - unless a trivial self-evident change such as fixing a typo. You can use the format Fixes #nnnn in your description to cause GitHub to automatically close the issue(s) when your PR is merged.
• Your change description explains what the change does, why you chose your approach, and anything else that reviewers should know.
• You have included any necessary tests in the same PR.

#5993

What does this PR do

This PR includes a source generator which generates SweepableEstimator class for all built-in estimators that defined in trainer-estimator.json and transformer-estimator.json

One example class looks like (kindly notice that the generated class is a partial class, and we manually code the rest of this class)

internal partial class LightGbmBinary : SweepableEstimator<LgbmOption>
{
        public LightGbmBinary(LgbmOption defaultOption, SearchSpace<LgbmOption> searchSpace = null)
        {
            this.TParameter = defaultOption;
            this.SearchSpace = searchSpace;
            this.EstimatorType = EstimatorType.LightGbmBinary;
        }

        internal LightGbmBinary()
        {
            this.EstimatorType = EstimatorType.LightGbmBinary;
            this.TParameter = new LgbmOption();
        }
    
        internal override IEnumerable<string> CSharpUsingStatements 
        {
            get => new string[] {"using Microsoft.ML;", "using Microsoft.ML.Trainers;", "using Microsoft.ML.Trainers.LightGbm;", };
        }

        internal override IEnumerable<string> NugetDependencies
        {
            get => new string[] {"Microsoft.ML", "Microsoft.ML.LightGbm", };
        }

        internal override string FunctionName 
        {
            get => "BinaryClassification.Trainers.LightGbm";
        }
}

The other half, which is manually coded, looks like this

internal partial class LightGbmBinary
{
        public override IEstimator<ITransformer> BuildFromOption(MLContext context, LgbmOption param)
        {
            var option = new LightGbmBinaryTrainer.Options()
            {
                NumberOfLeaves = param.NumberOfLeaves,
                NumberOfIterations = param.NumberOfTrees,
                MinimumExampleCountPerLeaf = param.MinimumExampleCountPerLeaf,
                LearningRate = param.LearningRate,
                NumberOfThreads = AutoMlUtils.GetNumberOfThreadFromEnvrionment(),
                LabelColumnName = param.LabelColumnName,
                FeatureColumnName = param.FeatureColumnName,
                ExampleWeightColumnName = param.ExampleWeightColumnName,
                Booster = new GradientBooster.Options()
                {
                    SubsampleFraction = param.SubsampleFraction,
                    FeatureFraction = param.FeatureFraction,
                    L1Regularization = param.L1Regularization,
                    L2Regularization = param.L2Regularization,
                },
                MaximumBinCountPerFeature = param.MaximumBinCountPerFeature,
            };

            return context.BinaryClassification.Trainers.LightGbm(option);
        }
}

Why create this bunch of SweepableEstimator classes instead of just having a factory class for all estimators

This is because of

• Avoiding having a giant factory class.
• the generated SweepableEstimator class includes useful information for modelbuilder to reconstruct source code training pipeline ( such as UsingStatement/NugetDependency )

Will those generated classes be internal only

That's the plan, external users will have to use CreateSweepableEstimator if they want to create a sweepable estimator.

[LittleLittleCloud]

@JakeRadMSFT

This will be essentially how user create a sweepable pipeline for multi-classification

[codecov]

Codecov Report

Merging #6125 (f3bc5ae) into main (24c4e5c) will increase coverage by 0.01%.
The diff coverage is 91.13%.

@@            Coverage Diff             @@
##             main    #6125      +/-   ##
==========================================
+ Coverage   68.29%   68.30%   +0.01%     
==========================================
  Files        1092     1095       +3     
  Lines      241554   241706     +152     
  Branches    25150    25152       +2     
==========================================
+ Hits       164974   165107     +133     
- Misses      70030    70046      +16     
- Partials     6550     6553       +3     

Flag	Coverage Δ
Debug	68.30% <91.13%> (+0.01%)	⬆️
production	62.81% <100.00%> (-0.01%)	⬇️
test	88.77% <90.78%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...rosoft.ML.SearchSpace/Attribute/OptionAttribute.cs	100.00% <ø> (ø)
...rc/Microsoft.ML.SearchSpace/Option/ChoiceOption.cs	94.44% <ø> (ø)
src/Microsoft.ML.SearchSpace/Option/OptionBase.cs	100.00% <ø> (ø)
...soft.ML.SearchSpace/Option/UniformNumericOption.cs	96.20% <ø> (ø)
src/Microsoft.ML.SearchSpace/SearchSpace.cs	64.63% <ø> (ø)
....ML.AutoML.Tests/Utils/DoubleToDecimalConverter.cs	50.00% <50.00%> (ø)
...t.ML.AutoML.Tests/Utils/FloatToDecimalConverter.cs	50.00% <50.00%> (ø)
....ML.AutoML.Tests/SweepableEstimatorPipelineTest.cs	95.34% <93.75%> (-4.66%)	⬇️
...icrosoft.ML.AutoML.Tests/SweepableExtensionTest.cs	94.73% <94.73%> (ø)
...ML.SearchSpace/Attribute/BooleanChoiceAttribute.cs	55.55% <100.00%> (ø)
... and 7 more