Have just joined up to say pretty much the same thing as arnow, though I’ve experienced the problem with far more than just one website. I appreciate the fact that you’ve tried to resolve this as quickly as possible, but unfortunately your new Markdownifier isn’t anywhere even close to being as good as Instapaper.
I tend to clip a lot of newspapers, and I could always rely on Instapaper to do a remarkably good job of separating the wheat from the chaff, while also providing essential information about the article, such as the name of the author, the date, and the publication it appeared in. Markdownifier comes close on the last of these, but gives the website, not the name of the publication. Instapaper tends to be able to accurately recognise and include the main image for a particular story, whereas Markdownifier often seems to throw it out. I’ve also seen some very odd results with captions. As an example, “(Video: Jenny Starrs/Photo: Jabin Botsford/The Washington Post)” became “(Jenny Starrs/The Washington Post)” though neither the video nor the photo survived. With a page from the Sydney Morning Herald, I’ve seen Markdownifier half digest and regurgitate a video panel that had nothing to do with the main story. It removed any reference to the relevant part of the panel, but included two photos from each of the eight advertised stories, with the two photos being at different sizes.
There’s also the question of style. In the past I’ve always tended to use Web Archives because the page that Instapaper produces tends to be far more readable than any of the other options. Though it’s far from being the only issue, column width is a significant factor here. Instapaper typically produces columns of about 75 characters including spaces (a little less than I’d prefer, but absolutely fine); Markdownifier is producing columns that are more like 175 characters wide, which is way outside any readability guideline that I’ve ever seen.
As things stand, I’m reduced to trying to persevere with Instapaper, via arnow’s workaround, but with mixed results. So from my point of view, two weeks after the problem first came to light, there’s currently no workable solution.
I hope you won’t mind me pointing out that Instapaper have obviously been working on this sort of thing for quite a while now, and the best and easiest way forward is probably going to be to find some way to get Instapaper working with DevonThink again.
Perhaps it’s time to contact them and see if you can’t work out some sort of deal.