refactor: rename struct_conversion benchmark to complex_type_conversion

- Rename benchmark file to better reflect its scope (struct, list, map)
- Fix incorrect comment: "native shuffle" -> "JVM shuffle"

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: andygrove

native/core/Cargo.toml

@@ -133,7 +133,7 @@ name = "parquet_decode"
 harness = false
 
 [[bench]]
-name = "struct_conversion"
+name = "complex_type_conversion"
 harness = false
 
 [[bench]]

native/core/benches/struct_conversion.rs …/core/benches/complex_type_conversion.rsnative/core/benches/struct_conversion.rs renamed to native/core/benches/complex_type_conversion.rs native/core/benches/struct_conversion.rs renamed to native/core/benches/complex_type_conversion.rs

@@ -15,10 +15,11 @@
 // specific language governing permissions and limitations
 // under the License.
 
-//! Benchmark for struct column processing in native shuffle.
+//! Benchmarks for complex type processing in JVM shuffle row-to-columnar conversion.
 //!
 //! This benchmark measures the performance of converting Spark UnsafeRow
-//! with struct columns to Arrow arrays.
+//! with complex type columns (struct, list, map) to Arrow arrays via
+//! `process_sorted_row_partition()`.
 
 use arrow::datatypes::{DataType, Field, Fields};
 use comet::execution::shuffle::row::{