Fix indentation and closing-paren placement for multi-resource try-with-resources

[saloni-eng]

Fixes #1392.

Try-with-resources statements with more than one resource were
formatting inconsistently: the first resource stayed glued to try (,
and the closing ) was glued to the last resource, e.g.:

try (var input = Files.newInputStream(Path.of("./input"));
    var output = Files.newOutputStream(Path.of("./output")); ) {

This changes multi-resource try statements to break after (, put
each resource on its own line, and place the closing ) back at the
same indentation as try, matching how other multi-line constructs
(e.g. argument lists) are formatted:

try (
    var input = Files.newInputStream(Path.of("./input"));
    var output = Files.newOutputStream(Path.of("./output"))) {

Single-resource try-with-resources statements are unaffected.

Changes

• JavaInputAstVisitor.visitTry: conditionally open a plusFour
  indent and force a break after ( only when there's more than one
  resource; close the indent level before the trailing break so the
  closing ) renders at the outer indent instead of the resource
  indent.
• FormatterTest: added multivariableTryWithResources covering the
  new formatting.
• testdata/B21465217.output, testdata/B26159561.output: regenerated
  golden files that encoded the old (pre-fix) multi-resource layout.

Testing

mvn clean test passes, including the CR/LF/CRLF idempotency checks
in FormatterIntegrationTest.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[25f2005869-glitch]

This fix is directionally correct, but I’d strongly prefer keeping this PR focused on the try-with-resources formatting issue only. The re-indentation churn in FormatterTest.java is much broader than the actual behavior change and makes the patch harder to review and reason about.

Could you revert the unrelated formatting-only edits and keep just the targeted logic change plus the relevant regression/golden updates? That would make the fix easier to validate and reduce the risk of accidental behavioral changes.