strip body on head request

[tzilist]

Hopefully fixes issue #45. Not sure exactly where this logic belongs but seems to make sense to me. :)

[tirr-c]

The response header of HEAD should be identical to that of GET, but it is tricky because of Content-Length header. It is handled by Payload trait from Hyper, which is implemented for hyper::Body.

If you give empty body to the response, Content-Length will be 0, which leads to the mismatch of headers. We need to implement Payload manually, but it requires Poll from futures 0.1 (from function signature).

[tzilist]

Oh wow, that is a great point! I'll dig into this a bit more then. :)

[tzilist]

Holding off on this until questions in #45 are resolved

[aturon]

Given how #45 has gone, I'm gonna close this out. It'd be great to get a PR documenting Hyper's behavior and how Tide relies on it, though!