toUnicode fails for malformed input

[sidonath]

Hi there,

Thanks for developing and maintaining this package 🙇

Earlier today, I ran into a weird error when another script produced a malformed string which is not a punycode, but looks like one: "xn--http.com".

Per RFC 3490, toUnicode should never fail:

   ToUnicode never fails.  If any step fails, then the original input
   sequence is returned immediately in that step.

Unfortunately, this is not always the case with punycode.toUnicode:

> punycode.toUnicode('xn--http')
RangeError: Invalid input
    at error (punycode.js:42:8)
    at decode (punycode.js:235:5)
    at punycode.js:389:6
    at map (punycode.js:57:20)
    at mapDomain (punycode.js:84:18)
    at Object.toUnicode (punycode.js:387:9)

It would be great if punycode silently ignored exceptions and returned the given input.