Regression in incomplete_tag

[Keno]

On JuliaSyntax master:

julia> s = "CompletionFoo.test(1, 1, "
"CompletionFoo.test(1, 1, "

julia> Base.incomplete_tag(Meta.parse(s, raise=false, depwarn=false))
:none

vs Base:

julia> Base.incomplete_tag(Meta.parse(s, raise=false, depwarn=false))
:other

I looked into adjusting the tag selector a bit, but couldn't come up with a good version that also distinguished [x (with space), [x) and ( and ( , so there may need to be some adjustment to the source-location representation itself. @c42f can you take a look?

[Keno]

ref JuliaLang/julia#56110

[c42f]

It appears this regression happened in #397

[c42f]

@Keno I think this is fixed by #518. I'm not sure I understand all the distinct test cases you were trying to cover - can you have a quick look and see whether I need to add more? Thanks!