io.circe.pointer.literal.PointerInterpolatorSuite seems Flaky (or there's a bug?)

[MartinHH]

The CI for PR #2156 failed with:

io.circe.pointer.literal.PointerInterpolatorSuite:
  + The pointer string interpolater should parse valid absolute JSON pointers 0.04s
  + The pointer string interpolater should parse valid relative JSON pointers 0.002s
  + The pointer string interpolater should work with interpolated values that need escaping 0.0s
==> X io.circe.pointer.literal.PointerInterpolatorSuite.The pointer string interpolater should work with arbitrary interpolated strings  0.294s munit.FailException: /home/runner/work/circe/circe/modules/pointer-literal/src/test/scala/io/circe/pointer/literal/PointerInterpolatorSuite.scala:47
46:
47:  property("The pointer string interpolater should work with arbitrary interpolated strings") {
48:    Prop.forAll(ScalaCheckInstances.genPointerReferenceString) { (v: String) =>

Failing seed: ti47UAvFZKoH5ggwYNhAsvoTmWb6j9CsslvYR17sYqM=
You can reproduce this failure by adding the following override to your suite:

  override def scalaCheckInitialSeed = "ti47UAvFZKoH5ggwYNhAsvoTmWb6j9CsslvYR17sYqM="

Falsified after 28 passed tests.
> Labels of failing property: 
Expected Right(/foo/~0/bar) but got Left(io.circe.pointer.PointerSyntaxError: Syntax error at 5; expected: token character)
> ARG_0: "~"
> ARG_0_ORIGINAL: "⠾Ḏᩁ鈋㚞ठۙ닂똈ᤂ貖ᖸâ㩍埗繤휅♅蟻⤬~0ᖐ弳叅અ㋕"

That can be reproduced on the main (series/0.14.x) branch.

[MartinHH]

Ah, seems like a known issue - quoting from #1991 (comment) :

Note, there is probably still more to do here in the future. The tests are slightly different in their structure for Scala 2/3 and from glancing at the RFC I suspect there are issues with the current pointer implementation when a reference contains Unicode characters which are represented by surrogate pairs of bytes under UTF-8.

[armanbilge]

ping @isomarcte via #1998 (comment)