Add support for Microsoft-specific modifiers

[MathieuNls]

Microsoft-specific modifiers are keywords can be used to modify declarators to form derived types.

The specification for the modifiers themselves is here:
https://docs.microsoft.com/en-us/cpp/cpp/microsoft-specific-modifiers?view=vs-2019.

If you are OK with these changes, I'll go ahead and open a synchronized PR on the cpp grammar to add the Microsoft modifiers there and fix the error that this PR introduces:

tree-sitter-cpp# ./tree-sitter-linux-x64 generate

Unresolved conflict for symbol sequence:

  attribute_specifier  •  'extern'  …

Possible interpretations:

  1:  (_declaration_specifiers_repeat1  attribute_specifier)  •  'extern'  …
  2:  (constructor_or_destructor_definition_repeat1  attribute_specifier)  •  'extern'  …

Possible resolutions:

  1:  Specify a higher precedence in `_declaration_specifiers_repeat1` than in the other rules.
  2:  Specify a higher precedence in `constructor_or_destructor_definition_repeat1` than in the other rules.
  3:  Add a conflict for these rules: `_declaration_specifiers`, `constructor_or_destructor_definition`

Fixes #41

[calixteman]

@maxbrunsfeld ping

[maxbrunsfeld]

This is very well done, and thanks for the reference link and the thorough tests. I left some feedback which is related to code size optimization and naming.

[MathieuNls]

@maxbrunsfeld ; I've addressed the comments and opened tree-sitter/tree-sitter-cpp#79 to reflect the changes on the cpp project.

Note that the test on tree-sitter/tree-sitter-cpp#79 won't pass until this is closed as it relies on the 0.17.0 version that I propose here.

[MathieuNls]

@maxbrunsfeld ping

[maxbrunsfeld]

Thanks so much; great work on this. I will try to look at the C++ PR later today.

[MathieuNls]

The last thing required for tree-sitter/tree-sitter-cpp#79 tests to pass is to tag 88592f1 as 0.16.1