Split Source and Package Parser (Part 2)

[fabianhjr]

This has two commits, the first one refactors SourceToken a bit and the second one implements an (unpolished) Package Lexer/Parser and refactors Idris/Package.idr to use this new Lexer/Parser.

Feedback is appreciated. First part is #89.

One pro is that package names can now have dashes in their name UwU

Cleanup of SourceToken

There are many organizational and consistency changes, one of the more important ones being Literal being rename StringLit to correspond with CharLit, IntegerLit, DoubleLit and extracting more common code to Parser/Lexer/Common.

Change to ParseError tok and ParseError

ParseError tok was renamed ParsingError tok and ParseError was refactored to ParseError tok to accept a token type.

[fabianhjr]

Hi @ziman, thanks for the feedback, those two points have been added to this merge request.

[ziman]

Great, thanks! I've got some more questions, I'll put them as comments in the diff.

[ziman]

Somehow I can't see why the lexer works, but the CI symbol is green so I'm probably missing something.

[fabianhjr]

Hey @ziman, I think this is way cleaner now, thanks for the feedback.

[ziman]

I've got just a few naming suggestions, and maybe a proposal to merge Ident and DotSepIdent.

Otherwise I think it's good to go.