updating pruning func for varlingam for faster processing

[svenpieper]

The recent update replaces a previous loop structure with a more efficient use of advanced indexing to fill the variables obj and exp. This improves code readability and may speed up execution, especially with large datasets.

Instead of iterating through each element block in blocks, a 3D array structure is now created to hold flattened ancestor indices for each block. This is done by extracting the ancestor indices for each block using blocks[:, 0, ancestor_indexes] and shaping them into a flat structure ancestor_indexes_flat.

Populating the obj variable is also optimized using advanced indexing, directly copying blocks[:, 0, i] into obj for more efficient assignment.

For the exp variable, advanced indexing is used similarly. Flattened ancestor indices are inserted into exp[:, :causal_order_no], and remaining elements from blocks[:, 1:].reshape(len(blocks), -1) are placed into the corresponding position in exp[:, causal_order_no:]. This update improves code clarity and enhance execution speed drastically.

During my research I noticed that the pruning takes a very long time and this happens especially with longer data sets with many nodes, this should fix this issue. Atleast for Varlingam.

[kunwuz]

Thanks so much! It looks great! :)