feat: Add Parquet Export Support

[BingqingLyu]

Feature

Add Parquet file export support to NeuG's COPY TO command, enabling users to export query results to high-performance Parquet files for analytics and data lake integration.

User Scenarios

1. Export for external analytics: Data engineers can export Cypher query results to Parquet files for analysis with Spark, DuckDB, or Presto
2. Compression optimization: Configure compression settings (none, snappy, zlib, zstd) to balance file size and performance
3. Large dataset export: Export millions of rows without memory issues using streaming writes

Requirements

P1: Core Parquet Export

• Basic export with default SNAPPY compression
• Streaming batch writes (no OOM for large datasets)
• Type mapping (NeuG types to Arrow/Parquet types)
• Arrow schema metadata preservation

P2: Compression and Performance Options

• Configurable compression: none, snappy, zlib, zstd
• Configurable row group size
• Dictionary encoding control

P3: Complex Data Type Support

• Vertex/edge object serialization
• List/array property handling
• Date and timestamp serialization

Syntax

Basic export (default SNAPPY):
COPY (MATCH (n:person) RETURN n.*) TO 'person.parquet';

With options:
COPY (MATCH (n:person) RETURN n.*) TO 'person.parquet' (compression='zstd', row_group_size=65536);