Replace markdown grammar with maintained version

[wooorm]

Description

This replaces the grammar used to highlight markdown with a maintained one.

I’ll add more sample files later. First wanted to bring this to your attention.

Related to: GH-6339.

Checklist:

• I am changing the source of a syntax highlighting grammar
  • Old: https://github.com/atom/language-gfm
  • New: https://github.com/wooorm/markdown-tm-language

---

Important: the current scope is source.gfm. The new scope is text.md.
There seems to be some Go code to handle text.html.markdown and text.md, which I changed.
The reasons why I prefer changing source.gfm to text.md is that:
a) markdown is closer to a text, markup, language, than to a source language that is “evaluated” or so
b) markdown isn’t a superset of HTML, it has its own complex rules for that, so text.html.markdown isn’t great IMO
c) Since it was (probably) added, GFM now bases itself on CommonMark (CM), bringing it much closer to normal markdown (for example, fenced code blocks are now “normal” markdown). This grammar, other than GFM features, also includes some other extensions (frontmatter, GitHub-specific features, directives, math), which aren’t in GFM

But, I wanted to bring this to attention because:
a) it’s still early days, I can be convinced to use something else
b) source.gfm is used in other existing grammars.

Notably, I see these grammars that use source.gfm:

• jawee/language-blade (not updated in 2 years)
• atom/language-ruby (archived)
• language-viml (@Alhadis)
• wooorm/markdown-tm-language (that’s me, I can do that!)

I imagine @Alhadis to be able to update this, and I can raise an issue with jawee/language-blade, although it’s not actively maintained, so I’m not sure that can be solved.

[Alhadis]

I prefer changing source.gfm to text.md is that:
a) markdown is closer to a text, markup, language, than to a source language that is “evaluated” or so

Correct. A language's type field is a good indicator of what prefix a TextMate grammar's scope should use: programming and data-type languages should use source.*-scoped grammars, while markup and prose-type languages should use text.*-scoped grammars instead.

It's unclear why the Atom team chose source.gfm for their markdown grammar, but it—along with virtually every other choice of scope-names they used—is an objectively poor choice of scopes that conflicts with established naming conventions.

a) it’s still early days, I can be convinced to use something else

I started a clean rewrite of a markdown grammar several years ago, but never published it (as I was reluctant to share a crappy 2-hour project stub with the world…). I'm sure I'll finish it eventually: I simply wanted to indent code-blocks without screwing up the rest of the document's highlighting:

b) source.gfm is used in other existing grammars.

Atom packages are likely to reference this scope for compatibility with the editor's default Markdown package (source.gfm), while others may support source.gfm and text.md in tandem, whichever the user has available.

Notably, I see these grammars that use source.gfm:
[…] language-viml (@Alhadis)

Only in VimL's injectionSelector field, which does nothing on GitHub because injections only affect grammars bundled in the same package/submodule (a wise restriction, IMHO). Ergo, it's use in language-viml is benign and can be ignored. 👍

[lildude]

LGTM. Thanks.

Note: this PR will not be merged until close to when the next release is made. See here for more details.