Normalize HTML attributes

[Crissov]

It would be helpful if Babelmark canonically reordered generated HTML attributes so that implementations that only differed in this regard were shown collectively.

Example

![foo](bar "baz")

Most implementations generate

<img src="bar" alt="foo" title="baz">

but Rdiscount and Pandoc put title before alt:

<img src="bar" title="baz" alt="foo">

and Python-Markdown puts alt first.

<img alt="foo" src="bar" title="baz">

[xoofx]

Indeed. Don't know when I can work on this but I will keep it around. thanks

[Crissov]

Add to this: ignore custom data-* attributes (e. g. Github) and possibly ids and classes (e.g. headings and footnotes).

[Crissov]

I've looked into the JS to find out where you normalize the HTML until I finally figured out that this is done on the babelmark-proxy server-side already, i.e. the JSON returned by the Azure server, e.g. for https://babelmark.azurewebsites.net/api/get?text=foo, already includes an html_safe string for each queried service.

NUglify does not seem to support reordering attributes or removing data-*.