Mozilla bug 1995614: Atomize ATTR_TYPE strings and any single digit strings during tokenization. r=hsivonen

Add static atoms for "2" -> "9", and do a fast path check for any single digit strings during tokenization and return the static atom directly.  Also atomize type attributes during tokenization.  This patch also cleans up some code in Element::SetAttr and nsHtml5TreeOperation so that it's easier to pass through an atom.

Differential Revision: https://phabricator.services.mozilla.com/D269859

Author: dpalmeiro

src/nu/validator/htmlparser/impl/Tokenizer.java

@@ -998,8 +998,21 @@ private void emitOrAppendCharRefBuf(int returnState) throws SAXException {
      * @return the buffer as a string
      */
     @Inline protected String strBufToString() {
+        // CPPONLY: String digitAtom = TryAtomizeForSingleDigit();
+        // CPPONLY: if (digitAtom) {
+        // CPPONLY:   return digitAtom;
+        // CPPONLY: }
+        // CPPONLY:
+        // CPPONLY: boolean maybeAtomize = false;
+        // CPPONLY: if (!newAttributesEachTime) {
+        // CPPONLY:   if (attributeName == AttributeName.CLASS ||
+        // CPPONLY:       attributeName == AttributeName.TYPE) {
+        // CPPONLY:     maybeAtomize = true;
+        // CPPONLY:   }
+        // CPPONLY: }
+        // CPPONLY:
         String str = Portability.newStringFromBuffer(strBuf, 0, strBufLen
-            // CPPONLY: , tokenHandler, !newAttributesEachTime && attributeName == AttributeName.CLASS
+            // CPPONLY: , tokenHandler, maybeAtomize
         );
         clearStrBufAfterUse();
         return str;