[Feature] Add Native Scan Support for Apache Hudi Copy-On-Write Tables

[slfan1989]

Description

Add native scan support for Apache Hudi Copy-On-Write (COW) tables using Auron's vectorized execution engine. This enhancement enables Auron to accelerate Hudi table queries by converting FileSourceScanExec operations on Hudi tables to native Parquet/ORC scan implementations.

Scope

Supported Features

• COW (Copy-On-Write) tables: Native scan for Parquet and ORC base files
• Configuration switch: spark.auron.enable.hudi.scan (default: true)
• Timestamp handling: Automatic fallback when native timestamp scanning is disabled

Limitations (Initial Version)

• MOR (Merge-On-Read) tables: Not supported, automatically falls back to Spark
• Time travel queries: Falls back to Spark to preserve metadata semantics
• Spark version: Only Spark 3.0–3.5 (Spark 4.x not supported)
• Hudi version: Only Hudi 0.15.0