Potentially rework CSS parsing

[facelessuser]

I'm considering a rewrite of the CSS parsing approach. We may take the approach of parsing the CSS as returning something like an AST. Then we would take that AST and make decisions about what is valid and what is invalid and construct our selector objects.

This could have the potential to make some of the internals more straightforward and make expanding the syntax easier. This is not really critical, but something to consider. As we have many tests in place, as long as we are able to produce the same results, we should be good.