So one possible solution is to just add an option to never decompress
response bodies, but that's a bit of a hack. The real question is:
what's the correct way to handle these tarballs? Web browsers seem to
know not to decompress them, but AFAICT http-enumerator is correctly
following the spec by decompressing. Another possibility is to
"Seem to" is pretty much correct; it took years for some browsers to reliably handle them correctly. (Anyone else remember Mozilla saving compressed tarballs uncompressed?)