[AURON #2030] Add Native Scan Support for Apache Hudi Copy-On-Write Tables.

[slfan1989]

Which issue does this PR close?

Closes #2030

Rationale for this change

This PR adds native scan support for Hudi Copy-On-Write (COW) tables, enabling Auron to accelerate Hudi table reads by converting FileSourceScanExec operations to native Parquet/ORC scan implementations.

What changes are included in this PR?

1. New Module: thirdparty/auron-hudi

• HudiConvertProvider: Implements AuronConvertProvider SPI to intercept and convert Hudi FileSourceScanExec to native scans

  • Detects Hudi file formats (HoodieParquetFileFormat, HoodieOrcFileFormat)
  • Converts to NativeParquetScanExec or NativeOrcScanExec
  • Handles timestamp fallback logic automatically

• HudiScanSupport: Core detection and validation logic

  • File format recognition with NewHoodie* format rejection
  • Table type resolution via multi-source metadata fallback:
    • Options → Catalog → .hoodie/hoodie.properties
  • MOR table detection and rejection
  • Time travel query detection (via as.of.instant, as.of.timestamp options)
  • FileIndex class hierarchy verification

2. Configuration

• Added spark.auron.enable.hudi.scan config option (default: true)
• Respects existing Parquet/ORC timestamp scanning configurations
• Runtime Spark version validation (3.0–3.5 only)

3. Build & Integration

• Maven: New profile hudi-0.15 with enforcer rules

  • Validates hudiEnabled=true property
  • Restricts Spark to 3.0–3.5
  • Pins Hudi version to 0.15.0

• Build Script: Enhanced auron-build.sh

  • Added --hudi <VERSION> parameter
  • Version compatibility validation
  • Auto-enables hudiEnabled property

• CI/CD: New workflow .github/workflows/hudi.yml

  • Matrix testing: Spark 3.0–3.5 × JDK 8/17/21 × Scala 2.12
  • Independent Hudi test pipeline

Are there any user-facing changes?

New Configuration Option

// Enable Hudi native scan (enabled by default)
spark.conf.set("spark.auron.enable.hudi.scan", "true")

How was this patch tested?

Add Junit Test.

[slfan1989]

Spark 3.0 and 3.1 don’t support time travel yet, so I’ll revise the unit tests accordingly.