Markdownlint should instead run after Liquid parsing

A lot of the markdownlint functionality is hampered because of Liquid tags

• MD037 (Spaces inside emphasis markers) is disabled because of it
  • Our usage of underscores in path names for {% link %}s makes this rule parse incorrectly
• MD044 (Names should have correct capitalization)
  • Same issue as with MD037, it recognizes Liquid tags as explicit text.
• Reference links that are pointing to Liquid tags are not checked for link validity
  • E.g. [My link] with [Mylink]: {% link _sections.Tablets.md %}

While some of these issues are less of an issue with our HTML linting in place, there are other issues that cannot easily be worked around.

The best solution is as the title proposes, to extract the Markdown documents after the Liquid tags have been parsed so that markdownlint can correctly parse the documents as it expects. This is likely achieved by outputting the formatted documents as part of the Jekyll build process.

Start by tracing the Jekyll build process and the current markdownlint invocation, since no specific files or tests are named. Check how formatted documents could be emitted after Liquid parsing, then verify that markdownlint handles the listed emphasis, capitalization, and Liquid link cases correctly.