Fixed nested/spanned node issues

[lkspencer]

Fixes a number of issues with nested/spanned nodes and simplifies the regular expressions.

[lkspencer]

If you go to the CKEditor demo and paste the following HTML into the source, previously it would not find and replace all of the misspelled words. I've used several situations of spanned nodes and partial words and particularly teste followed by stem cell would previously get concatinated in the old code to testestem cell which has teste in there twice making it more difficult to properly find the right instance and correct it. This update takes care of every one of these situations.

<p>this is a test</p>

<p>I need to <strike>k</strike><em><strike>n<strong>o</strong></strike>o</em>w</p>

<p>if it will woork</p>

<p>testi knoow <strike>wo<u>o</u><em>r</em><strong>k</strong></strike></p>

<p>testily</p>

<p>teste</p>

<p>stem cell</p>

[artem-frolov]

\b is suitable for Latin characters but it doesn't work correctly with all Unicode characters.