Improve content type detection

[ameshkov]

For some browsers (for instance, for Chromium and FF) we could use "Accept" header to detect request content type.

For instance, here is an image loading with <img> tag:
Accept: image/webp,image/*,*/*;q=0.8

Does not work <script> tags.

Works for <link> in case if type is specified:
Accept:text/css,*/*;q=0.1

Document:
Accept:text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8

We can use Origin header as marker for xmlhttprequest.

If Accept starts with application/json, it is most likely XHR as well:
application/json, text/javascript, */*; q=0.01