Core: Migrate switch statements to switch expressions

[nssalian]

Continues #15494 (@mxm) and #15176 (@manuzhang)

Summary

Converts statement-form switch blocks in iceberg-core to expression-form (arrow syntax). Eliminates StatementSwitchToExpressionSwitch ErrorProne warnings and removes the risk of accidental fall-through in the converted code.

Rationale for the change

Per the [DISCUSS] Switching to new switches dev list thread, consensus was option 2
(merge sweeping syntax changes shortly before the release). Targeting 1.12.

Changes

• iceberg-core only. Pattern-matching for instanceof - I'll create a separate follow-up PR.
• The diff was generated by ErrorProne's auto-patcher (-XepPatchChecks:StatementSwitchToExpressionSwitch -XepPatchLocation:IN_PLACE), then spotlessApply, plus a small manual cleanup in PrimitiveWrapper.sizeInBytes() to keep trailing comments inline.
• SingleValueParser.fromJson was annotated @SuppressWarnings("MethodLength") because the arrow-form blocks push it past the 150-line checkstyle limit. Should be refactored later but out of scope for a mechanical syntax migration.

Test

• locally ran the core tests

[kevinjqliu]

I think its a good idea to add '-Xep:StatementSwitchToExpressionSwitch:ERROR', to baseline.gradle so that we can enforce this going forward and verify that all refactors are caught.

tested locally and found more places where we should refactor

[nssalian]

I think its a good idea to add '-Xep:StatementSwitchToExpressionSwitch:ERROR', to baseline.gradle so that we can enforce this going forward and verify that all refactors are caught.

tested locally and found more places where we should refactor

Makes sense. Did you find it in core or other modules? I kept this scoped to core for now. Adding that would scope to other modules. Happy to do a follow up if that works better.

[steveloughran]

commented. The new layout really makes a difference on assigment/return statements; elsewhere it can be a bit more verbose, but eliminates the risk of accident fall-throughs.

just think how many accidental fall-through bugs have been written since K & R decided that switch statements in C would fall through by default. And how many are still out there, in code we all depend on?

[nastra]

this should be good to go once the remaining comments have been addressed

[nastra]

LGTM. Are you planning to open PR(s) to address other modules and to enable '-Xep:StatementSwitchToExpressionSwitch:ERROR'?

[nssalian]

@nastra thanks for the review. Yes, I'll open follow up PRs for that and the instance of changes.

[nssalian]

From the discussion in the community sync up, any similar changes going forward should be in individual PRs and would be reviewer's discretion to allow those in.