add unique transformation

[Edwardvaneechoud]

This pull request adds a new deduplication feature to the LazyFrame API, allowing users to efficiently remove duplicate rows from data frames. It introduces both hash-based and streaming deduplication methods, supports deduplication on a subset of columns or entire rows, and ensures correct integration with the query optimizer and column pruning logic. Comprehensive tests and documentation are included to demonstrate and validate the new functionality.

Deduplication Feature:

• Added a new unique method to LazyFrame and TypedLazyFrame, supporting deduplication on specified columns or entire rows, with an optional sorted parameter for optimized streaming deduplication when input is pre-sorted. [1] [2]
• Implemented two new plan nodes, UniqueNode (hash-based deduplication) and SortedUniqueNode (streaming deduplication for sorted input), with appropriate logic for execution, schema propagation, and plan explanation.

Query Optimization and Column Pruning:

• Updated the optimizer to correctly handle deduplication nodes, ensuring filter pushdown and column pruning work as expected with the new deduplication operations. [1] [2]

Documentation and Testing:

• Added a new "Deduplicating" section to the documentation (docs/operations.md) with usage examples for the new unique method.
• Added comprehensive tests for deduplication, including subset deduplication, full-row deduplication, multi-column deduplication, sorted vs hash deduplication, error handling, filter pushdown, and column pruning.