Add more comprehensive tessts for A-labels in the "hostname" format

[jdesrosiers]

Since draft-07, the hostname format has specified that it includes IDNA2008 A-labels and that all valid hostnames should also be valid idn-hostnames. That means that it's not enough for hostnames to be valid punycode, they also need to be able to decode to a valid U-label including all the complex rules involved. This PR adds an equivalent A-label test for all of the U-label tests from the idn-hostname format tests to the hostname format tests.

I also did a little clean up. I found several tests issues including a few duplicate tests, incorrect test descriptions, and few other issues. I also reordered some of the tests so they were grouped logically. I did the clean in a separate commit to make review a little easier.

[jviotti]

I'm not an expert on hostnames in general, but at least what I see seems to make sense to me

[karenetheridge]

You've brought in restrictions from RFC5891 (e.g. "contains "--" in the 3rd and 4th position" -> invalid) which I don't think are correct. The hostname format is defined as a union of the requirements from RFC1123 and RFC5891, not an intersection, and "--" is a valid substring under RFC1123 as far as I can tell?

[jdesrosiers]

@karenetheridge The spec says,

Note that all strings valid against the "hostname" attribute are also valid against the "idn-hostname" attribute. Section 7.3.3

That means that any label in a hostname that starts with xn-- must be a valid IDN A-label. Otherwise it wouldn't be a valid idn-hostname.

The hostname format is defined as a union of the requirements from RFC1123 and RFC5891, not an intersection

I don't think that interpretation makes sense. A-labels (labels produced using the Punycode algorithm) are always valid RFC 1123 labels. So, that requirement would be meaningless if there were no requirement to validate those A-labels. That and the language that all hostnames are idn-hostnames is pretty convincing to me that hostnames need to be validated as ASCII-only idn-hostnames. I don't think it was a good idea to define it that way, but I think that's what the spec currently requires.