[selectors-4] Canonicalize into extlang form prior to :lang() matching

Fixes #4154

Author: frivoal

selectors-4/Overview.bs

@@ -24,7 +24,7 @@ Abstract: Selectors Level 4 describes the selectors that already exist in [[!SEL
 At Risk: the column combinator
 At Risk: the '':read-write'' pseudo-class
 Ignored Terms: function token, Document, DocumentFragment, math, h1, shadow tree, querySelector(), quirks mode, button, a, span, object, p, div, q, area, link, label, input, html, em, li, ol, pre, CSS Value Definition Syntax
-Ignored Vars: identifier, extended filtering, i
+Ignored Vars: identifier, i
 </pre>
 <pre class=link-defaults>
 spec:css-syntax-3; type:dfn; text:identifier
@@ -1925,8 +1925,12 @@ The Language Pseudo-class: '':lang()''</h3>
 
 	The element's <a>content language</a> matches a <a>language range</a> if
 	its <a>content language</a>, as represented in BCP 47 syntax,
-	matches the given <a>language range</a> in an <var>extended filtering</var>
-	operation per [[RFC4647]] <cite>Matching of Language Tags</cite> (section 3.3.2).
+	matches the given <a>language range</a> in an <i>extended filtering</i>
+	operation per [[!RFC4647]] <cite>Matching of Language Tags</cite> (section 3.3.2).
+	Both the [=content language=] and the [=language range=]
+	must be <i>canonicalized</i>
+	and converted to <i>extlang form</i> as per section 4.5 of [[!RFC5646]]
+	prior to the <i>extended filtering</i> operation.
 	The matching is performed case-insensitively within the ASCII range.
 	The <a>language range</a> does not need to be a valid language code to
 	perform this comparison.