Remap operator tokens to K"Identifier" during parsing

[c42f]

Having various operator token kinds is relevant during parsing, but they should potentially all be remapped to K"Identifier" afterward because their special status as operators is no longer meaningful once they live in the parse tree - they are essentially just identifiers which will be looked up with the usual scoping rules, just like any other identifier.

Code which processes syntax trees often has to do things like is_operator(x) || kind(x) == K"Identifier". Mapping them to identifiers simplifies things and fixes this issue. For the same reasons, I've also wished for this in JuliaLowering.jl.

One caveat is that this change might be unhelpful for the JuliaSyntax.tokenize() API - need to think about what to do there. (Perhaps operator kinds should be preserved in that case?)

If we do this, we should also do it systematically. For example, to fix https://github.com/JuliaLang/JuliaSyntax.jl/pull/456/files#r1692517031

[fredrikekre]

Code which processes syntax trees often has to do things like is_operator(x) || kind(x) == K"Identifier".

Ha, good timing for this issue. I commited fredrikekre/Runic.jl@9043278 just a day ago :)

[c42f]

@Keno notes in https://github.com/JuliaLang/JuliaSyntax.jl/pull/456/files#r1693352829

It also seems odd that e.g. isa gets turned into Identifiers, but + does not.

Yes this seems inconsistent.