Proposal: parallelize in the compare method for improved efficiency

[hlbotterman]

Description
The current implementation of the split processing is done sequentially, which can result in significant execution time, especially when dealing with large datasets or numerous splits or numerous imputers. In some cases, this sequential approach creates bottlenecks, leading to longer overall processing times than necessary.

Proposal
I propose introducing parallel processing for handling splits to improve efficiency and reduce the time required for the procedure. By utilizing joblib, we could take advantage of multi-core systems, allowing the splits to be processed concurrently.