LibWeb: Make Node.textContent more spec compliant

The current implementation felt a bit ad-hoc and notably allowed
textContent to operate on all node types. It also only returned the
child text content of the Node instead of the descendant text content.

Userland/Libraries/LibWeb/DOM/CharacterData.h

@@ -25,8 +25,6 @@ public:
 
     unsigned length() const { return m_data.length(); }
 
-    virtual String text_content() const override { return m_data; }
-
 protected:
     explicit CharacterData(Document&, NodeType, const String&);