Support more characters directly before negation

[jfmengels]

Follow up to #271

I went through https://lucamug.github.io/elm-cheat-sheet/elm-cheat-sheet.color.pdf and found a bunch more characters that should be allowed before -, which I've tested in the REPL. I've written tests for those. I've also added a few regressions tests for patterns that should not be supported (mostly things like <|- which the compiler understands as an unknown operator).

I have not succeeded in supporting all of them though. I'd love help with this.

[lue-bird]

in theory this one should not parse if we want to be that strict

-- UNKNOWN OPERATOR ------------------------------------------------------- REPL

I do not recognize the (==-) operator.

1|   a==-x
      ^^^
Is there an `import` and `exposing` entry for it? Maybe you want (==) or (-)
instead?

---

Also, the fact that the let...in-a test fails is bad. We probably either need

• to check on "n" that the previous is "i" and the one before isn't alphanumeric
• a different kind of negation parser for after in specifically.

Currently I have passing tests with

if
    (String.slice (offset - 3) (offset - 2) source == "i")
        && Basics.not
            (String.all Char.Extra.isLatinAlphaNumOrUnderscoreFast
                (String.slice (offset - 4) (offset - 3) source)
            )
then
    negationAfterMinus

else
    ParserFast.problem

[jfmengels]

Oops, right.

And I agree that the current system is not enough for in.

[jfmengels]

@lue-bird I don't think I'll merge this unless tests pass. Feel free to push to this branch (or create a new branch/PR on top of this one) if you'd like to continue the work.