Adding new _compatibility test_ `EnsoCompilerTest` to verify the new Rust based parser can produce the same `IR` as the original `AST` based one. The simplest way to execute the test from an empty repository is:
```bash
enso$ sbt bootstrap
enso$ sbt "testOnly *EnsoCompilerTest"
```

There are [GitHub Actions run](https://github.com/enso-org/enso/actions/runs/3087664644/jobs/4993266212#step:9:5187) on Linux as well as [run on Windows](https://github.com/enso-org/enso/actions/runs/3087664644/jobs/4993266370#step:9:5254) that show `EnsoCompilerTest` is being executed by the CI (good, as that means `.so` was properly built and linked to the JVM running the test). The [linux](https://github.com/enso-org/enso/actions/runs/3087664644/jobs/4993266212#step:9:5187) as well as [windows](https://github.com/enso-org/enso/actions/runs/3087664644/jobs/4993266370#step:9:5254) runs also demonstrate that failures in the `EnsoCompilerTest` suite fail the CI.

# Important Notes
Right now [there are five test failures](https://github.com/enso-org/enso/actions/runs/3087664644/jobs/4993266212#step:9:5187) - waiting for @kazcw to make sure `codeRepr()` doesn't contain spaces. However, as this PR is more about the infrastructure, I am disabling the currently failing tests in [031169b](https://github.com/enso-org/enso/pull/3723/commits/031169bd055d21e4b1a6049f2ad361adaae8c7c7)