Fix a bug where if we used fetch url with include_headers=FALSE and then read() happens to read in the headers, the trailing \r\n\r\n and then one byte of the body (but no more) then we accidentally dropped the one byte off the start of the body.