Removing old Ruby builds like Ruby 1.9-2.2 and JRuby 9.1-9.3 and TruffleRuby 19-22

Based on the excludes in https://github.com/ruby/setup-ruby/blob/afda160499b3ea79ea34ed7dc1191c3f68294309/.github/workflows/test.yml#L38
I'm thinking to remove old Rubies such as:

• 1.9. which is only built on ubuntu-22.04 and macos-latest but fails to build on all other platforms. It probably doesn't work well either and there are tons of exclude for it.
• 2.0, 2.1 and 2.2 since 2.2 is known to segfault (https://github.com/ruby/setup-ruby/issues/496) and likely none of them work well. Also 2.2 has this weird version mismatch explained in the README. Also the oldest version that builds on arm64 is 2.3.5 and it seems nice to have approximately the same versions on x64 and arm64 available (when possible).
• JRuby 9.1, 9.2 and 9.3, so we'd keep only supported release trains of 9.4 and 10.0.
• TruffleRuby releases which are EOL and older than 2 years, like everything older than 23.0.0 (which is compat 3.1, 22.3 is compat 3.0).

I want to keep @v1 and avoid jumping to @v2 to avoid unnecessary churn and confusion. The idea is to not maintain those anymore, if there was a @v2, @v1 would become "dead" and I don't want that and all the issue spam that would result from it.

I'm thinking to first remove support for them in setup-ruby (so it would result in an error then, or maybe at first a warning), and then after some time remove them in https://github.com/ruby/ruby-builder/releases/tag/toolcache.

Is anyone using those versions and have a working build they can link?
That would make me consider keeping some of these versions longer.

Examinez les exclusions de versions dans .github/workflows/test.yml et les chemins de prise en charge de setup-ruby, puis comparez-les avec la version du toolcache de ruby-builder décrite dans l’issue. Confirmez si les versions de MRI, JRuby et TruffleRuby listées disposent encore de builds fonctionnels ou d’utilisateurs ; le travail est terminé lorsqu’un plan de suppression ou de conservation est arrêté sans casser l’interface @v1.