Add requests/urllib3 work-around for intercepting gzipped bytes.

[dhermes]

Fixes #34.

/cc @tseaver @mfschwartz

@jonparrott Do you think we should cc the requests / urllib3 maintainers? I'm also working on a PR to fix the failing system test, but it is not going to be part of this PR.

[theacodes]

Go for it.

[mfschwartz]

I would suggest that rather than just Cc'ing the requests / urllib3 maintainers, we recommend adding a feature to their libraries so the caller can get ahold of the pre-decoded content to do things like computing checksums. I think that capability has more general value than just for this library.

[dhermes]

@Lukasa @shazow We're curious what you think of this hack.

For Google Cloud Storage (and likely other media APIs with checksumming) we "need" simultaneous access to the (gzip) compressed and uncompressed bytes so we can compute the checksum of the gzipped content (which is what is actually stored in "the cloud") while streaming the uncompressed bytes to a stream for the user.

[mfschwartz]

Opened #37 about the deflate encoding issue noted above.

[shazow]

@dhermes Looks like you did what you could with the API we have. :) It really does feel like the decoder stuff should be a registry, doesn't it?