fix: ensure scala 3 dirs use scala3 dialect for scalafmt

[ckipp01]

I think this is the cause of the error we're seeing in the Scala
Steward
where the scala213 dialect has issues with the Scala 3 code. This small
changes just adds in some fileOverrides to make sure the scala-3 dirs
set the runner.dialect as scala3 instead of scala213.

[ckipp01]

Odd, ./bin/scalafmt --check fails in CI, but locally when running .bin/scalafmt it doesn't format anything... Also does that get dynamically updated in CI or something? I'm not really sure why my change would impact this though 🤔 .

[bjaglin]

Also does that get dynamically updated in CI or something?

🤯 #1784 (comment) (run just before scalafmt) does mutate the faulty project/plugins.sbt. I'll open a PR to run it last (it makes sense anyway as scalafmt is faster to run and more likely to fail).

2 things remain mysterious to me:

1. Locally, dogfoodScalafixInterfaces' session save does not touch project/plugins.sbt, why does it do it on CI?
2. Why did the --test picks up the missing LF in this PR (and this PR only)? We might just be witnessing (1) in CI in most cases?

[bjaglin]

Regarding the original trigger for this PR: we are actually back to green since yesterday.

Failing runs (and failing runs only) have "ERROR Steward scalacenter/bloop-config failed", so I assume the problem is related to that. It's comforting that the output triggered by scala-3 files during steward's scalafmt run is not fatal, as scalafmt --test does return successfully in scalafix CI (modulo the weird dogfoodScalafixInterfaces case, which does not happen with steward anyway).

  org.scalasteward.core.io.process$ProcessFailedException: '"SBT_OPTS=-Xmx2048m -Xss8m -XX:MaxMetaspaceSize=512m" mill -i -p /home/runner/scala-steward/workspace/repos/scalacenter/bloop-config/scala-steward.sc show org.scalasteward.mill.plugin.StewardPlugin/extractDeps' exited with code 1.
   Cannot resolve -p. Try `mill resolve _` to see what's available.

I am not sure what changed in bloop-config/steward on July 6th though...

[ckipp01]

Failing runs (and failing runs only) have "ERROR Steward scalacenter/bloop-config failed", so I assume the problem is related to that. It's comforting that the output triggered by scala-3 files during steward's scalafmt run is not fatal, as scalafmt --test does return successfully in scalafix CI (modulo the weird dogfoodScalafixInterfaces case, which does not happen with steward anyway).

Yup, I noticed this as well. This is because a new release of steward came out with a fix for the actual error we were seeing. There still is an error for this one, but it doesn't fail the run.

[bjaglin]

🙇