README.md is detected as text/x-matlab

This Pandoc-flavoured markdown file

$ cat README.md
% Title
% Author

# This is a markdown file...

...with a title block for [Pandoc-flavoured markdown](https://pandoc.org/MANUAL.html#extension-pandoc_title_block)

is detected as text/x-matlab, not as text/markdown (as its suffix implies) or text/x-readme (as its prefix implies):

$ xdg-mime query filetype README.md
text/x-matlab

I think the reason is the following. Globs don't give a unique result since both README.* from text/x-readme and *.md from text/markdown match. Then xdg-mime looks at the file contents. Neither text/markdown nor text/x-readme define any magics. However text/x-matlab has this magic defined:

    <magic priority="10">
      <match type="string" value="%" offset="0"/>
    </magic>

Thus, the % as the first character makes xdg-mime thinks it is a text/x-matlab file.

I think this file should be identified by xdg-mime as text/markdown. The % magic from text/x-matlab clashes with pandoc requiring the % to be the first character for its markdown title block parsing.

I guess this could be worked around by adding a magic for the % character to text/markdown similar to that for text/x-matlab? I would be willing to provide a patch to do this, but I'm not sure if this is the right approach. Maybe xdg-mime should behave differently in such cases where the mime type from file contents (text/x-matlab) is not one of those from the globs (text/markdown and text/x-readme)? I.e. maybe xdg-mime should just return the highest-weight glob match (text/markdown) and ignore the file content matches in such cases?

Fwiw, I think this is a similar issue as #127 (closed).