chore(migrate): remove version check

[ematipico]

Summary

This PR fixes a number of issues in our migration rules:

• Remove the TheVersion service. It's not needed anymore. Less code :)
• Remove the check of rome.json inside the migration. We removed it, and we don't handle it anymore. Hence, the field configration_directory_path isn't needed anymore.
• Fixed the style rules migration, in particular the logic inside the rule move type.
  • It doesn't remove existing members anymore.
  • It always adds the correct number of separators, so it won't emit incorrect CST (this was causing the infinite loop)
  • I removed the useWhile from the styling rules because it's handled by use_while.rs. This was causing another infinite loop where useWhile and styleRules were creating one action after the other... lol!
• The migration command now formats the code at the very end. I'm surprised it didn't 😅

Test Plan

Our biome.json was causing the loop, so I added it as a test case in our testing suite.

I also run the migration to our biome.json and made sure there wasn't a loop anymore

[ematipico]

The first implementation was adding the new member at the very end. While it's more efficient, it basically moves around an existing member, causing plenty of diffs.

I decided that it's best to pay a bit more in memory, but at least we do swap the members in the same position, so the number the final diff is very limited.

[ematipico]

@dyc3 @Conaclos

Here's an update. The RuleMover project I had in mind was a failure. It was becoming too complex and it was full of all sort of bugs. I fixed enough bugs to make it working for the displayRules, and I removed its usage in the other rules.

I ran the PR against the ecosystem and found three projects that were failing. Hence, I also added their configuration files in the migrate_v2.rs test file. This should cover most of the cases, and we should be able to catch possible loops early enough.

Apologies for the problems caused, but the migration should be stable now :)

[Conaclos]

Thanks for the update!

Hence, I also added their configuration files in the migrate_v2.rs test file

Have you tried to reduce to a minimal reproduction fore every failure?

[ematipico]

Yes, the failures were due to RuleMover, which I removed from the rules. I then updated the single migrations against regressions.

We required bigger configs to test all the migrations together. However, the majority of the issues were around the move of rules