Hi,

I read your code and congrats, there are very interesting tricks used here to make your diff algorithm.

I have also worked on diffing a while ago, and found that the most efficient (in term of size and execution speed) seemed to be Myer's diff algorithm: https://blog.robertelder.org/diff-algorithm/

I minified and golfed it and reached a size of 406 bytes (without gzip):
http://xem.github.io/miniDiff/

So this is not really an issue, but if you're intersted to make your code shorter, this could be an idea.

Cheers,
Max