wget doesn't follow GitHub redirects, GitHub returns HTTP 302 redirects for raw file URLs. The wget -O- command doesn't follow redirects by default, so it downloads the redirect HTML page instead of the actual file. When piped to tar, this fails because tar receives HTML instead of a gzip archive.
Replace wget -O- with curl -L or wget -O- --max-redirect=5 would also work, but curl is more concise.