Investigate extending standard MIME implementation to handle incomplete MIME parts, especially: * Incomplete file metadata * Unknown character encoding * Unknown content type (using "magic number" detection?) * Double-encoded base64/quoted-printable binary