Cache HTTP redirects

[njsmith]

When you GET a file at archive.org using its canonical URL, then they always issue a 302 Moved Temporarily redirect to some internal URL where the data is actually stored. When using the ZS library to access such a file, following all these redirects works fine, but it doubles the number of round trips. We should cache the redirection, and make future requests directly to the internal URL.