Taking the format that @aaronpk has used for aaronpk/IndieLogin.com.
Also make sure the redirect array gets cleared at the first request,
just in case the transport object is reused.

Rebuild the debug property from the raw headers passed to the new
CURLOPT_HEADERFUNCTION and the response body. This will match the raw
HTTP response more than previously. It will also contain all the
redirecting steps and HTTP status lines.

CURLOPT_HEADERFUNCTION is the recommended method for collecting the
response headers. Because every request in the redirect chain contains
headers (at the minimum a location line), we can track which URL’s
headers are being parsed and collect a history of URLs.

* added docs for how to use the library
* fixed setting the user agent for requests
* add option to send http headers for HEAD requests