The tokenizer configuration has become difficult to handle due to the additional manual transliteration rules. Allow to have a separate rule file that is given to the ICU library as is.
