Improve image extraction

There’s a couple sites that lazy load images or substitute custom tags for img tags by executing scripts.
An example of lazy loaded images can be found here if you disable scripts and visit the page, upon inspection they use a custom attribute called data-src 川崎 遺体遺棄事件 被害女性の通夜 親族や友人など死を悼む | NHK | 事件

As inoreader is able to extract the images I believe they either use custom extractors per site or they got better heuristics, from checking NewsBlur’s code it looks like you just select img tags and pray src is the right attribute.
A “simple” solution may be to make it configurable/allow adding additional tags/attributes per feed which shouldn’t be too resource intensive either.
I may be able to contribute a pull request if you think the idea would be ok to implement but not worth the effort.

(As a sidenote the encoding of this page gets all sorts of messed up if you use the original or story tabs)

Upon further investigation postlight/parser might be causing issues too as that is what extracts the actual content.