Additional validation of Content-Length parsing was introduced in bf90f3a however, the value is not striped and therefore a value of '0 ' causes a ValueError.

This is a particular issue because if Content-Length is the last header in a request and parse_line is being called then the \r\n of the end of the header is interpreted as a multi-line continuation and appends the space to the end in HTTPHeaders.parse_line: new_part = " " + line.lstrip().

@bdarnell