Add a regression test for url_search_params::sort taken from WPT

[npaun]

While working on integrating Web Platform Tests as a test suite for Cloudflare's worked, I discovered a situation where ada-url's implementation of url_search_params::sort appears not to be following the URL spec.

The URL spec provides the following requirements for URLSearchParams.sort:

The sort() method steps are:

1. Sort all tuples in this’s list, if any, by their names. Sorting must be done by comparison of code units. The relative order between tuples with equal names must be preserved.

— https://url.spec.whatwg.org/#example-searchparams-sort

WPT's urlsearchparams-sort.any.js test uses the following example to demonstrate this:

{
"input": "ffi&🌈", // 🌈 > code point, but < code unit because two code units
"output": [["🌈", ""], ["ffi", ""]]
},
— https://github.com/web-platform-tests/wpt/blob/master/url/urlsearchparams-sort.any.js#L11

So the spec requires "comparsion of code units" but as far as I can tell, it isn't very clear about what encoding should be used to perform the comparsion. Based on the WPT test, and checking other implementations like Node and Chromium, the encoding used is UTF-16. This makes sense, as this matches the behaviour of Javascript's String type.

I can fix this by implementing something like ada::idna::utf8_to_utf32 to do a conversion from utf8 to utf16 before sorting the keys, but please let me know if you have an alternative approach in mind - I'd be happy to implement it that way instead.

[lemire]

@npaun Sorting in UTF-32 and UTF-8 should be equivalent.

Regarding UTF-16, it will differ from UTF-32/UTF-8 but only when surrogate pairs are involved (I believe this is correct).

We can avoid conversion and memory allocation.

[lemire]

I am going to close this PR and reopen another one with a fix.