Web entities translated to graphical representation in Rich Text, original entity lost

HTML Entities entered in Wiki Markup are being lost when edited in Rich Text mode. This is breaking other Wiki Markup statements.

(For info on entities, see http://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references)

1) Go to a page, click Edit, click Wiki Markup
2) Enter the following text:

*one » two » three*

3) Click Preview, see the pretty bold words and right angle-quotes
4) Click Wiki Markup, see the same text as in 2
5) Click Preview see the same pretty display as in 3
6) Click Wiki Markup, see the same text as in 2 (again)
7) Click Rich Text, see the pretty bold words and right angle-quotes (again)
8) Click Wiki Markup, see that the Entities are gone, replaced with actual right angle-quotes
9) Click Preview, see that the bold text isn't and that the asterisks are now visible
10) Click Wiki Markup, see the same text as in 8
11) Click Rich Text, see the same text as in 8 (and 10)
12) Click Wiki Markup, see that the asterisks have now been escaped (i.e. "slash-asterisk")
13) Click any tab you like, the text never again changes from this point

The code generated by the entity-less markup displays the actual illegal-in-HTML characters instead of the originally entered legal entities

One note: I just tried the above with the following text:

*one* *»* *two* *»* *three*

The entities still get replaced with right angle-quotes, but the bolding continues to work and the text is unchanged after that. So while this seems to "work" (but creates illegal HTML), that much individual manipulation shouldn't be necessary, should it?