Major improvement to Text compatibility

I’ve just updated the service to include a major improvement for Text-mode page compatibility.

Before, if the service’s text parser couldn’t properly parse the document for whatever reason (often due to malformed HTML), it would completely fail.

Now, there’s a fallback parser that’s much simpler and filters out much less, but is far more likely to work with malformed pages or unexpected conditions.

(If you see a Text version of an article without the black Instapaper navigation bars on top and bottom, it’s using the fallback parser.)