Didn't you say that doxygen can parse markdown as well?

It can parse it, yes. But it generates automatically pages for every new markdown file (named the name of the markdown file). We don't want this here, but just a module documentation.

In detail: when adding a doc.md instead of a doc.txt (and adding doc.md to the FILE_PATTERN config var of doxygen), the page tree of doxygen looks like:

▼ RIOT Documentation
   doc                       # containing an empty page
   RIOT in a nutshell
   Contribute to RIOT
   The quickest start
   Structure
   Todo List
 ▶ Deprecated List
 ▼ Modules
  ▶ Boards
  ▶ CPU
  ▶ Drivers
  ▶ Kernel
  ▶ Networking
    Packages
  ▶ System
▶ Data Structures
▶ Files

I don't want this empty doc page ;-)

I see.

A somewhat related configuration option I came across by accident:
See http://stackoverflow.com/questions/13368350/use-the-readme-md-file-as-main-page-in-doxygen and
https://github.com/authmillenon/RIOT/blob/d50de568d1c00162900d9ba478f67790b1b10091/doc/doxygen/riot.doxyfile#L924
http://superuser.com/questions/685817/doxygen-specify-a-markdown-file-as-the-main-page

I saw this, too. But since PRs like #4285 will introduce further pages, I thought it to be less consistent. Additionally, the doxygen markdown still requires labels to be of any use to doxygen (Sub-sections in page tree, TOC, and ability to reference sections or pages from within doxygen). This isn't parsed however by GitHub properly (as well as the @ref references to other pages). Have a look at the main page on GitHub vs the intended result to see what I mean.

Rebased to current master

ACK and go!