LibWeb: Throw out decoded UTF-32 data in HTMLTokenizer after parser runs

This ends up saving quite a bit of memory on many pages, since UTF-32
uses 4 bytes per code points.

As an example, it reduces the footprint on https://gymgrossisten.com/
by 2 MiB.

Libraries/LibWeb/HTML/Parser/HTMLTokenizer.h

@@ -145,6 +145,8 @@ public:
     // This permanently cuts off the tokenizer input stream.
     void abort() { m_aborted = true; }
 
+    void parser_did_run(Badge<HTMLParser>);
+
 private:
     void skip(size_t count);
     Optional<u32> next_code_point(StopAtInsertionPoint);