Grammar Improvements

[NeuralCoder3]

https://anydsl.github.io/thorin2/langref.html

The grammar seems to contain mistakes:

• The rule I for index literal is never used
• pack and array expressions contain i but this does not mean import here
• v is used but never declared
• The precedence rules seem to be not complete
• s+ in g is confusing as s is optional X?+ = X*
• The precedence rules list Π Sym : e does this refer to .Pi Sym ( : etype )? , edom n? (can be confused with Π b → ecodom) => unicodes should be used consistently
• I think more rules are unused that I did not catch

[NeuralCoder3]

Maybe a good solution would be to write the grammar as tree-sitter (https://github.com/NeuralCoder3/tree-sitter-thorin)
and generate the document from it.

The tree-sitter grammar can be automatically tested (e.g. to parse the lit tests) and allows additional uses such as

• syntax highlighting in IDEs
• intelligent auto-completion
• linting
• code folding
• auto-formatting
• find variable declaration/usages (structure navigation)