LUCENE-2207: CJKTokenizer generates tokens with incorrect offsets LUCENE-2219: Chinese, SmartChinese, Wikipedia tokenizers generate incorrect offsets, test end() in BaseTokenStreamTestCase