[docs] Add array type support documentation for Lance integration

[XuQianJin-Stars]

Purpose

Linked issue: Addresses user questions about array type support in Lance integration (community feedback)

This PR adds comprehensive documentation for using array types with Lance lakehouse storage, particularly focusing on vector embedding scenarios which are critical for machine learning and AI applications.

Brief change log

• Updated website/docs/streaming-lakehouse/integrate-data-lakes/lance.md:
  • Added ARRAY<t> to Lance data type mapping table
  • Added new "Array Type Support" section with detailed documentation
  • Documented array type use cases (vector embeddings, multi-dimensional features, time series data)
  • Provided complete SQL examples for creating tables with array columns
  • Included end-to-end vector embedding example for recommendation systems
  • Added Python examples demonstrating Lance vector similarity search capabilities
  • Documented best practices for working with array columns in Lance
  • Added performance considerations for large embeddings and vector workloads

Tests

This is a documentation-only change. No code tests are required.

Documentation can be verified by:

1. Building the documentation website locally: cd website && npm install && npm start
2. Reviewing the rendered markdown on GitHub
3. Checking that all code examples follow correct SQL and Python syntax

API and Format

This change does not affect any API or storage format. It only adds documentation for existing array type support in Lance integration.

Documentation

This change enhances existing documentation by providing comprehensive guidance on array type usage with Lance. It does not introduce a new feature but documents the existing capability of using ARRAY<t> data types with Lance lakehouse storage, which is particularly important for users working with vector embeddings and machine learning workloads.