Normalization of quote character leads to issues in both CMS and site rendering in combination with IE8

IE8 does not render "<p>foo &apos; bar<p>" as "foo ' bar" but as "foo &apos; bar". This is a known IE8 bug and only affects this character entity. Background: &apos; is not a HTML4 entity, but an XML/HTML5 one.

The fact that we only now are bitten by this IE8 bug (and not earlier) is because the way the ' character is normalized has changed in one of the 7.9 maintenance releases (as a result of the changes we made to both the HtmlCleaner and CKEditor). An overview of how 7.8, 7.9.1, 7.9.8 and trunk store the string
"<p>foo ' bar &apos; baz &#39; lux</p>"
when entering the text in source view and clicking save (see note below).

Bottom line: customers can have content with "&apos;" as literal text, entered either on purpose or because of conversions, leading to editing problems (opening and saving a document in IE8 transforms "&apos;" to "&amp;apos;" CMS-9479) and on the site customers only now notice the rendering problem in IE8 (CMS-9434).

Note that CKEditor also does cleanup when switching from source to wysiwyg mode, it actually replaces all &apos; and &#39; for ' but that does not help us (enough).