KMeans Init Sparsity Support

[md-shafiul-alam]

Add sparsity support to KMeans Init and fix a few bugs in daal sparse kmeans++ init, onedal kmeans++ init, and kmeans infer. Specific changes planned or made in this PR.

• Fix distance calculation for sparse data in daal KMeans++
• Allow oneDAL Kmeans++ init to take n_trials same as daal and scikit-learn
• Fix difference between daal Kmeans++ dense and sparse results
• Implement KMeans init sparse support for CPU (just calling daal implementation - cpu)
• Fix oneDAL KMeans sparse infer on GPU
• Update Kmeans infer for sparse data to allow result options same as dense

I have verified that

• Daal kmeans init results are same for sparse and dense data
• oneDAL kmeans init results are same for sparse and dense data
• oneDAL kmeans init results are same on cpu and gpu
  • Not same for dense data unless we compute initial centroids for dense GPU using cpu implementation

[samir-nasibli]

Thank you @md-shafiul-alam for thework done!

I see that only one test suit here updated. It worth to increase or maybe updates tests for better coverage of new fixes and features.

[Alexandr-Solovev]

In general LGTM, but we have to discuss optional results, due to looks like its not necessary with the latest comments, and it makes sense to remove/update it. Also, it would be nice if you add tests with direct comparison between results on the same data, but with different formats(dense , sparse) with some tolerance to track that it works fine

[md-shafiul-alam]

/azp run CI

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[md-shafiul-alam]

/intelci: run

[Alexandr-Solovev]

/intelci: run

[md-shafiul-alam]

/intelci: run

[md-shafiul-alam]

/intelci: run

[md-shafiul-alam]

/azp run CI

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[Vika-F]

LGTM

[md-shafiul-alam]

/intelci: run

[md-shafiul-alam]

The failures are not related.