------------------
- *FEATURE:* Character counts used instead of word counts where possible in
order to make the algorithm work well in the language independent
mode (without a stoplist) for languages where counting words is
not easy (Japanese, Chinese, Thai, etc).
- *BUG FIX:* More robust parsing of meta tags containing the information about
used charset.
- *BUG FIX:* Corrected decoding of HTML entities &128; to &159;