404 with Content-Encoding gzip

[Kevin Ryde <user42_kevin@xxxxxxxxxxxx>]

The Project Gutenberg 404 page is sent "Content-Encoding: gzip",

    https://www.gutenberg.org/no-such-page-exists

Current cvs emacs-w3m displays the raw gzip bytes.

Nosing around w3m-show-error-information, I see the intention is
w3m-decode-encoded-contents.  It reaches zlib-decompress-region, but
that returns nil.  Is that because it's a multibyte buffer?  The rest of
w3m-decode-encoded-contents spawning gzip works for me.