NB this is timeboxed to 5 hours
Figure out if we really create an improvement in diffs by introducing the new empty line treatment. To do so, please follow the process we established for changes in moved paragraphs, which means:
- generate random article samples
- check if they were improved / worsened/ unchanged (imagining we had implemented the whole feature, where line numbers are correct and emptly lines are seperate visual entities)
- document the results