Add to SDE theory algorithms for which a streaming version does not exist

[emanueledellavalle]

• Entropy & Mode: It uses a Fold operator that computes the local and global occurrence of each unique data, and then using a Map, it computes the entropy or retrieves the most common value. It is not a streaming operator since it has a worst-case space complexity of O(n) but can be distributed.
• Skew & Kurt: returns the skewness or kurtosis of the columns. They use an Iterare operator to compute the needed statistics in the first iteration and then compute the Skewness and Kurtosis with a second iteration. They need to perform two passes over the data so they are not streaming but can be distributed.
• Covariance & Pearson Corr.: returns the co-variance or the correlation of two columns. It drops the unneeded columns, and then, similar to the Skewness, it uses an Iterate to compute the needed statistics in the first iteration and the co-variance or the correlation with the second. It needs to perform two passes over the data so they are not streaming but can be distributed.