Core, Spark: Add JMH benchmarks for Variants

### Feature Request / Improvement

There's no JMH benchmarks for variants, so no way to see if behaviour scales well.

### Query engine

None

### Willingness to contribute

- [x] I can contribute this improvement/feature independently
- [ ] I would be willing to contribute this improvement/feature with guidance from the Iceberg community
- [ ] I cannot contribute this improvement/feature at this time

Start by locating the Variant implementation and any existing JMH benchmark entry points in the repository; compare how current benchmarks are structured. Add benchmarks for Variant behavior at relevant scales and verify that they produce usable results for evaluating performance.

Written by the indexing model from the issue text.