As discussed in #1327 a very simple formatting test is added to the native build.

It's a fairly basic smoke test, but I actually verified here EvaristeGalois11#1 that it's enough to catch problems like the recent bug 😁

Fixes #1331

COPYBARA_INTEGRATE_REVIEW=#1331 from EvaristeGalois11:native-test 04b4238
PiperOrigin-RevId: 866015724

PiperOrigin-RevId: 871704300

This was introduced via search-and-replace in d8216e8.

Fixes #1340

COPYBARA_INTEGRATE_REVIEW=#1340 from mernst:remove-duplicate-type-annotation 2bb3000
PiperOrigin-RevId: 874817120

After https://bugs.openjdk.org/browse/JDK-8268850, `VariableTree#getType` returns a new AST node `VarTypeTree` for variables declared with `var`.

The new node could be handled in the visitor by implementing `visitVarType`, but that API is only available on the latest JDK versions using var, which would require creating a new version-specific visitor implementation. To defer doing that, this fix re-orders the logic that handles variable types to check if the next token is `var` first, to avoid needing to handle the new AST node.

PiperOrigin-RevId: 876244813

I was debugging something where having a `toString` implementation in `TypeWithDims` would have been helpful.

PiperOrigin-RevId: 876329640

The plugin is failing to build after changes in [1] to add usages of records.

[1] 4962993

https://github.com/google/google-java-format/actions/runs/22568168992/job/65368984173

PiperOrigin-RevId: 877811635