feat: add SQL dialect support with Spark SQL dialect

[yuchen-pipi]

Summary

• Add SqlDialect enum (Default, Spark, PostgreSql, MySql, Sqlite) and SparkDialect implementation using DataFusion's unparser Dialect trait
• Refactor to_sql() to accept an optional dialect parameter instead of a separate method per dialect
• Add Python API support: query.to_sql(datasets, dialect="spark")

Spark SQL dialect differences

• Backtick identifier quoting
• STRING type instead of VARCHAR
• EXTRACT(field FROM expr) for date parts
• LENGTH() instead of CHARACTER_LENGTH()
• TIMESTAMP without timezone info
• Subqueries in FROM require aliases

Usage

Rust:

use lance_graph::{CypherQuery, SqlDialect};

let sql = query.to_sql(datasets, Some(SqlDialect::Spark)).await?;

Python:

sql = query.to_sql(datasets, dialect="spark")

Test plan

• 7 new Spark SQL integration tests (backtick quoting, filters, relationships, complex queries, dialect comparison, PostgreSQL dialect)
• 5 unit tests for SparkDialect trait implementation
• 12 existing to_sql tests updated and passing

🤖 Generated with Claude Code

[ChunxuTang]

The motivation for adding SQL dialects makes sense to me.
Left some comments on the design of the new feature.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 91.66667% with 7 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
crates/lance-graph/src/query.rs	74.07%	7 Missing ⚠️

📢 Thoughts on this report? Let us know!

[ChunxuTang]

@yuchen-pipi
LGTM! Just left a very minor comment.
Meanwhile, could you fix the style lint error?

[jja725]

@ChunxuTang The workflow passed, do you mind helping merge it since we don't have this button on our end.