Extract port

[john-kurkowski]

• Add optional port as the 4th member of the ExtractResult namedtuple
• Deemphasize documentation about rejoining the ExtractResult namedtuple members by dot
  • Since port in a URL-like string is optional and joined by colon, not dot

Fixes #272.

Compatibility

• This breaks the previously documented ".".join(extract_result) pattern
  • If callers aren't careful, the pattern will produce awkward strings like "foo.bar.com.None" or "foo.bar.com.8080", or raise a TypeError from trying to join str with Optional[int]
  • Arguably, relying on an ExtractResult being a namedtuple is historical. Still, this PR would complicate things for users relying on that join.

Other Options

1. Make port an attribute of ExtractResult but not a member of the tuple, like urllib.parse.urlsplit does
2. Don't merge this; callers can get the port using the standard library (and the port is validated)

   •   >>> import urllib.parse
       >>> print(urllib.parse.urlsplit("//foo.bar.com:8080").port)
       8080

   • As a consolation to Get the port from the provided URL to extract function #272, speed up this package when passed a urllib.parse.{ParseResult,SplitResult}: see Add method to extract from a urllib.parse.{ParseResult,SplitResult} #274

[john-kurkowski]

After thinking about it, this library is focused on domain names, not every component of a URL. I defer URL parsing to Python's standard lib.