Regex for matching URLs does not fully conform to RFC 3986 URI Generic Syntax

[ericherman]

In #1 it was noted that https://www.ietf.org/rfc/rfc3986.txt is not the same as what is being used in the url-check.py today.

In particular, some valid characters are only valid in parts of the URI, and the current regex does not distinguish this.
Also, some valid characters are not matched by the current regex.