v0.29
utf-8 output is more friendly to Linux
修改demo.py的代码格式，使得符合pep8规范
将tab换成空格，使用is判断对象是否为None
add a example of using sklearn+jieba
fix bug and change version
fix some bad cases
add one more example
adjust prob_trans for better support of name entity; fix some bad cases
improve POS tagging
hot fix of cut_all=True
in the speed test: initialize first to ignore the time of dict loading
make lazy load thread safe
update test case