Handle ERB blocks within HTML

[romaricpascal]

Current parser extension, based on the sample in Kramdown's documentation, ignores ERB when it's nested inside HTML tags. For example:

<<%= "div" %>></<%= "div">>
<div <%= "attribute='value'" %>></div>
<div >
  <%= "Erb test" %>
</div>

This can always be circumvented by using <%= tag.ELEMENT do %>, but could use sorting out. Only 3rd example feels actively useful, though (as for the other 2 tag.ELEMENT would provide a clearer code, I think).

Plan would likely be to override add_text to do some further scanning when adding a text node to try and detect one or several ERB tags, in a similar way that Kramdown's parse_raw_html does for HTML tags.