I've added another script to my project:
https://gitlab.com/pklat/reiso
its reorder.pm
it tries to find similar files, so that tar+7z later can provide best compression. the output is a file list which can be fed to tar with '--files-from=FILE'
this is done by compressing every two files!
since there could be many combinations, this is of course very very slow!!!
I might try to optimize it later.
please edit the script options first, if you wanna try it.
sorry, if this idea has already been used.
thanks.