SpellChecker: instead of parsing the line based on position, use Line…

… + Col approach - this way, if we have offset error, it will only affect the current line

and the offset error will not get carried over to the next line.

SpellChecker: use `UTF8Lengh()` method to calculate real length of tokens, this should fix problems when using non ASCII chars in the document
 and the spelling error marker does not correctly mark the word