I did, and, as far as I can tell, it's not helpful, because the HTML tags themselves are almost never meaningful in the pages we get back. For example, it's all well and good to be able to pick out the data between table tags, but then I still have to sort through the table data.
(I think the problem is that, in my case, it's the data and not the HTML tags that are significant - HTML::Parser is good for cases where the tags are the significant piece. If someone has used HTML::Parser in a similar way, please let me know.)
I wouldn't think that it will happen anytime soon, specially if it depends on the same people that make their web servers ignore HEAD requests, or kill the connection if you don't fake an "official" user-agent... Take Hotmail for an example.