[GUIDE] Interoperability with Polars, DuckDB, and PyArrow (Zero-Copy Analytics)

[FrancescAlted]

Blosc2 provides zero-copy Apache Arrow interoperability via CTable.to_arrow(), from_arrow(), and the PyCapsule protocol. Some data teams may want to leverage Blosc2 for high-ratio compressed columnar storage while running analytical queries using modern query engines like DuckDB and Polars.

Suggested Work

Create doc/guides/interop_arrow_duckdb_polars.md and link it from doc/guides/index.rst.

Provide runnable code examples showing:

1. PyArrow Round-trip: Converting a blosc2.CTable to a pyarrow.Table / RecordBatch and back with zero copying where supported.
2. Direct SQL queries with DuckDB: Querying Blosc2 tables using DuckDB's Arrow scanner:

   import duckdb
   import blosc2
   
   table = blosc2.open("dataset.b2z")
   res = duckdb.arrow(table.to_arrow()).filter("price > 100").aggregate("avg(score)").df()

3. Polars Integration: Ingesting a Blosc2 table into Polars (pl.from_arrow(table.to_arrow())) and converting Polars DataFrames into Blosc2 CTables.
4. Performance & Memory Comparisons: Illustrating memory usage during queries on compressed Blosc2 tables vs. raw in-memory DataFrames.