Is that ok?

Fixed issues mentioned in comments above.

Added git commands.

There are some other things that I'm still working on, so please don't push until I say that's all....

Added more information about force-push.
Changed the look of labels.
Added note to watch out what MR is one trying to make.

Fixed one more typo.

Fixed typos (thanks for pointing them out!).
Optimized png image.

Removed last line of debug

Removed last parts of debug.

I moved the conditional and add qAbs to make sure negative near-zero values are caught as well.

Move sanity checks to KisMarkerPainter

Another test file with instruction is in the bug report: https://bugs.kde.org/show_bug.cgi?id=405119

I fixed both Recent Documents and progress bars, but repeated error messages are probably still to be fixed. I couldn't reproduce it with a simple touched file on Linux. I'll try to find some mp4 files to check if the behaviour is different then (it shouldn't).

Fix multiple layer merging data loss

Actually it will probably break if the path from root to the putAfter node would be: N D D D N N D D D N (where D means "marked to be deleted" and N means otherwise). It won't happen with flattening, but since this function is used in other places, too, it's probably better to change it (to get the last ascendant instead of first), even though this case is a bit less common. (And the previous version of findPerfectParent() would cause the same data loss in this case).

Fix coding style in KoUnit

I removed all to[Unit] functions altogether, so now every time you want a conversion you need to create a KoUnit. I can change it though so the to[Unit] functions would use toUserValue functions.

As far as I know, it's not pushed yet. My credentials: