[VL] allow to read from CSV/Hive TEXT natively

### Description

CSV/TEXT format is a still commonly used some workloads(legacy workloads like HIVE job). Velox does not support CSV/TEXT yet and it may require big efforts on support this. Instead we could support this via arrow dataset(https://github.com/oap-project/gazelle_plugin/tree/main/arrow-data-source) and export the arrow columnar vector as Velox columnar vector after scan.

Start by reviewing the arrow-data-source implementation linked in the issue and how its scan output could connect to Velox vectors. Define the CSV/TEXT and Hive TEXT scope, then verify that scans can read these formats through Arrow and export usable Velox columnar vectors.

Written by the indexing model from the issue text.