[VL] Read Parquet file generated by parquet-thrift failed

### Backend

VL (Velox)

### Bug description

https://github.com/apache/gluten/actions/runs/23858922685/job/69563146517?pr=11860
```
2026-04-01T17:52:37.3856475Z Error Source: RUNTIME
2026-04-01T17:52:37.3856711Z Error Code: INVALID_STATE
2026-04-01T17:52:37.3857455Z Reason: Converted type VARCHAR is not allowed for requested type ROW,nestedstringcolumn:VARCHAR>
2026-04-01T17:52:37.3858333Z Retriable: False
2026-04-01T17:52:37.3859258Z Expression: !requestedType || isCompatible( requestedType, isRepeated, [](const TypePtr& type) { return type->kind() == TypeKind::VARCHAR; })
2026-04-01T17:52:37.3860821Z Context: Split Hive: file:///opt/shims/spark34/spark_home/sql/core/src/test/resources/test-data/parquet-thrift-compat.snappy.parquet 0 - 10550 Task Gluten_Stage_4_TID_4_VTID_73783
2026-04-01T17:52:37.3861558Z Function: convertType
2026-04-01T17:52:37.3861924Z File: /work/ep/build-velox/build/velox_ep/velox/dwio/parquet/reader/ParquetReader.cpp
```

### Gluten version

_No response_

### Spark version

None

### Spark configurations

_No response_

### System information

_No response_

### Relevant logs

```bash

```

Start with the linked GitHub Actions job and the parquet-thrift-compat.snappy.parquet resource named in the error. Read velox/dwio/parquet/reader/ParquetReader.cpp, focusing on convertType and the reported nested ROW conversion failure. Done means the Parquet file can be read without the INVALID_STATE error.

Written by the indexing model from the issue text.