Fixed Unicode string encoding issues for python 2/3.

[goerlitz]

As mentioned in #29 Unicode strings are not properly handled when using Python 2 and lead to exceptions.
This fix makes sure that Unicode string encoding works for Python 2 and Python 3 as well.

[TheJefe]

why not text.encode('utf-8') for python 2?

[goerlitz]

Short Explanation:

Python 3 uses unicode by default while python 2 differentiates between ASCII and unicode.

Therefore, if text.encode('utf-8') is applied on an (encoded) unicode string in python 2 it will result in
UnicodeDecodeError: 'ascii' codec can't decode byte ... in position 1: ordinal not in range(128).

Longer Explanation:

In python 3, type(u'Köln') and type('Köln') both give <class 'str'> and class 'str' is encoded to class 'bytes'.
In python 2, type(u'Köln') gives <type 'unicode'>, type('Köln') gives <type 'str'>, and type 'unicode' is encoded to type 'str'.

Since function annotate() requires input of type 'str' any unicode string in python 2 has to be passed as encoded 'str' and further calls of text.encode('utf-8') will result in UnicodeDecodeError.

Consequently, text.encode('utf-8') should only be called in python 3.