I have not actually set up the robots.txt in maxcdn. But the cdn is not indexing, which is what I am wanting, doing the site: search for the cdn shows no results. For the main site though the images are falling out of the index, even though there is a site map for them and they are still accessible from their normal url.

I am not using Wordpress, the site is using PrestaShop, so it does not have those plugins.

This is how it is set up.

cdn.site.com is a cname of www.site.com so images are accessible from www.site.com/image.jpg and cdn.site.com/image.jpg but when they are served from the cdn.site.com/image.jpg they have the canonical header that points to www.site.com/image.jpg I cannot understand why that would de-index all of the images on www.site.com though

It is a pull zone.

That is actually what has been done, I have seen the article before. But there are two issues, one they are not indexing and it has been a couple of weeks. But the more major issue to me is that using the cdn url none of the link juice from the main domain is being passed to the cdn sub domain. I am trying to figure out why Google is not respecting the canonical header for the images. It would seem to me, that according to what Matt Cutts says that it would. But it is not.

This one is stumping me and I need some help. I have a client who's site is www.site.com and we have set them up a CDN through Max CDN at cdn.site.com which is basically a cname to the www.site.com site. The images in the GWT for www.site.com are de-indexing rapidly and the images on cdn.site.com are not indexing.  In the Max CDN account I have the images from cdn.site.com sending a canonical header from www.site.com but that does not seem to help, they are all still de-indexing.

You are going to have to add a manual rewrite rule to the htaccess, preferably at the top. This might help, http://stackoverflow.com/questions/1321123/redirect-from-subdomain-to-domain-htaccess  But like Travis said, be careful, make a back up, because you can break your site doing this.

Can you access the pages at blog.yoursite.com? Do they show up in the browser? It could be an issue that has always been there that Google just found.

My view is a little bit different, but to be 100% I would have to look at the site. If the application / module is using style="display:none" to hide the tag, I would not use it. The reason being is that multiple h1 tags on a page dilute or in some peoples opinion cancel out other h1 tags. There are valid reasons to hide h1 tags, like when you are doing fade ins or have a slider with it in there. So I am of the opinion that even if it is hidden Google will pick up on it. So you do run the chance of being caught for keyword stuff or it could cancel out your h1 tag that is displaying. (As far as I know there is no clear 100% information on how Google handles multiple h1 tags. I have heard all sorts of things including it registers the first one seen in code, it registers the last one seen in code, it combines the weight given to all of the h1 tags, and even that they don't matter it is text size above the fold that matters)

If it were me this is what I would do. I would remove the text from the banners and use the built in functionality for the text. Depending on the location of them on the page really depends on how I would tag them, they could be h2, h3, span, ect. But I would opt for on page visible text over using a graphic with text. I realize that you do not like the style that it uses, but it is just a style that can usually be mimic'd with a style sheet. The reason I usually opt not to use text on banners is twofold. It cannot be translated and it cannot be searched.

I might have used a bad example with Rand, but it is amazing how many companies take paid posts or reviews. Places like Allure, Vogue, Huff Post, NY Times, ect. What you are really hitting is the demographic that thinks they are reading something that is impartial, but in reality they are just being advertised to under the guise of "News". I always make sure the link is nofollow, so I do not really consider it blackhat, be shakey marketing, maybe. But in the end it comes down to dollars and cents. I regularly have posts that are paid in the 500-1000usd range. When you first do it, it is a leap, because there is no SEO value at all. But the largest return I have had was a post that in 3 days grossed 50k in sales on high margin products. The posts usually die fast, because people want the latest greatest thing. But they end up getting shared and work for the most part generally.

Generally it is against Google's guidelines if you are buying a do follow link. So in doing something like this you need to weigh the SEO vs Marketing value. I personally do it with a lot of my clients, but not for SEO purposes, it is for marketing / sales purposes. My clients are generally e-commerce sites for a little insight. See some markets have blogs that people read every time it is posted. Kind of like a Rand Fishkin article in our market. If Rand came out tomorrow witha whiteboard friday and said this product is great, it boosted Moz's organic SEO by 100% think how many people would flock to buy it. Most of the time since paid links should be no follow, there is no SEO benefit, but at the end of the day if it puts more money in yours or your clients pocket than was spent, then it is worth it. I will say that I buy links, I make sure they are not followed and I buy them when I think they can create sales for my clients. As far as I know that model fits in the Google guidelines because buy making sure they are not followed I am not buying them for SEO purposes.

I would, personally I have disabled my tag page and author page since there are no tags, and only one author. As for the categories I have them no-indexed.