Add text() pseudo-attribute selector

[kzu]

Being able to filter or select nodes based on their text content is
something that is supported in XPath since its inception. CSS doesn't
allow this, but since we're in a bridge where everything runs as XPath
expressions in the end, it would be convenient to extend the CSS
syntax to allow for this.

Proposal is to simply allow text() as if it were an attribute name, in
the same context and with the same capabilities as if it were an attribute
value retrieval. This would allow it to also participate in all text matching
capabilities of the attribute selector.

For example:

• Get div with text foo: div[text()=foo]
• Get divs with text ending in suffix: div[text()$=suffix]

All the other matching strategies, like *=, ^= should also work as-is.