However I need one small improvement. If a file has, say, 100 000 lines and there's like 1000 different lines between current and previous version, it would be really beneficial to be able to hide 99 000 identical lines and see only differences. It would really make the life easier

And to be completely robust, there could be also "Add <n> context lines" as it is in standalone command line diff tools, so that just <n> identical lines around each change are displayed and rest is hidden.
Thanks in advance.