Wouldn’t the opposite be better? I’m not sure you could take advantage of the vast majority of files in the zip files being unchanged if you kept compressed archives.
Not sure what you mean but typically you don’t need to track changes of libraries to that level. At least not in the context of a repo using those libraries. I am thinking of treating them like binary .dlls.
Source control is really good at compressing text files as they evolve over time, but isn’t optimized to handle binary assets. Since a single-line patch changes the entire zip archive, you’d risk growing the size of the repository based on the number of patches.
