Original PR description

Checking the filelist size before checking for existence based on `NamedToInfo` is unnecessarily convoluted, we can just compare the list of expected filenames to the `namelist()`.

And using entry positions to open them is pretty much insane, just address the files by name, that's both clearer and more reliable. *Technically* for xlsx files we should probably go through `[Content_Types].xml` to retrieve the actual path to the shared strings file, but accessing that file directly is already an improvement I think.