nodewords_metatag_from_node_content() needs improving

There are problems in nodewords_metatag_from_node_content():

• Bug: The strip_tags() call is completely ignored.
• Task: The code is confused by use of $text when the $result variable already exists.
• Task: The _nodewords_teaser_match_callback() could be just merged inline.
• Feature: It doesn't IMHO make sense that the alt tags of an image be added inline, possibly within a sentence, there should at least be a way of customizing it so that e.g. it's wrapped in parenthesis.
• Task: In the interest of code simplicity, each filter call should be separate rather than stacked.
• Task: There's no way of customizing a global regex for filtering node teaser output, so there's no point in keeping the code in nodewords_metatag_from_node_content().
• Task: Move the php filter check higher up in the function to avoid wasting processing time.