There don't appear to be any floating images in the linked article, so I'm wondering if there is actually a bug in Calibre somewhere.
Anyway, here is a foolproof, but manual way to do this:
- Install the Firebug add-on to Firefox
- View the desired page in Firefox
- Open Firebug by clicking the little bug icon on the lower right corner of Firefox
- Now you need to click the 'Inspect' Icon in Firebug, it's the second icon from the left in Firebug's toolbar, the one with an arrow and a box
- Once that's clicked, hover over the different elements of the document, you want to hover over the element that covers all the text, you'll know you have the right one when you highlight the one that says <table....xxxx.....>
- Now, in the lower half of the window where firebug is showing the DOM, right click on the table element and click copy innerhtml
- Paste that into Notepad or some other text editor
- Save that as a new html file, then load that in FIrefox
- Save the simplified version of the doc in Firefox as a complete web page
- Load the new html file into Calibre
- Convert
You'll get output that looks like what I've attached.
The recipe framework can be used to do this all of this manual work automatically as I mentioned before. It's definitely a bit more work than a standard recipe because there isn't any rss feed, but it's doable - you could also create your own list of articles to use as the 'feed'.