I ported many more tests, pushed in new commits. Still a lot to go but we're getting closer.

I added a few more commits. It's not 100% ready yet but it's close so I'd appreciate a review.

I'll finish remaining pieces & address feedback when I'm back from vacation.

This is a lot to review. Honestly, I'm okay with merging when you feel it is ready and then we can increment in smaller PRs from there.

As discussed at the meeting today, I'll merge it soon (tomorrow or on Wednesday) unless someone wants to have a look.

Some final remarks/questions from me (you can review them instead if you want):

1. I updated selector-native slightly so that it passes some backported tests it used to fail - some Sizzle logic wasn't backported there.
2. Re test/unit/extending.js: it contains tests for custom pseudos which we'll soon drop so I haven't backported it
3. Re test/unit/utilities.js: it tests a few APIs that we already have tets for on the jQuery side; it also contains tests for Sizzle.attr, though. I assume they're not needed as jQuery attr tests should be enough.
4. Tests using Sizzle.matchesSelector have been migrated by replacing Sizzle.matchesSelector(selector, elem) by jQuery(elem).is(selector). Sizzle.matches tests weren't backported as it's a private API that no code in jQuery or Sizzle uses.
5. Lots of tests using Sizzle have been migrated to use jQuery.find - should we document that method? If not, should we migrate those tests to use jQuery? Some queries need to be modified then. jQuery.find accepts as many as four parameters; do we want to support them all? Some tests are using all of them right now.
6. Safari has surprisingly good support for CSS selectors that other browsers don't support. E.g. :not(complex selector) works in Safari. Therefore, I UA-sniff in tests & run some tests just in Safari OR if Sizzle is present (by checking for jQuery.find.compile as existing tests do).