[Bug Fix] Resolved load error due to difference in num_examples between configs

Added processing to remove rows with empty text.

Enabled config for with and without metadata such as headers, footers, quotes. Enabled direct download using sklearn

[Bug Fix] Corrected data file download

Initial Commit of the 20-Newsgroup dataset with groupname as labels and split into train, validation and test sets.