test(datafusion): cover SQL schema evolution reads

[liujiwen-up]

Purpose

Linked issue: #258

Add DataFusion SQL coverage for schema evolution tables generated from Java/Spark consistency fixtures, focusing on mixed-format reads through SQL. The rename coverage uses schema_evolution_rename_column, whose fixture contains parquet/orc/avro files.

Brief change log

• Add a small assert_sql helper for table-formatted SQL result assertions.
• Cover SQL reads for mixed-format add column and type promotion tables.
• Cover SQL reads for rename, drop, and reorder/move schema evolution cases with SELECT, WHERE, and ORDER BY id.

Tests

• cargo fmt --check: passed
• git diff --check: passed
• cargo test -p paimon-datafusion --test read_tables schema_evolution --no-run: passed
• cargo test -p paimon-datafusion --test read_tables schema_evolution -- --nocapture: not fully runnable locally because /tmp/paimon-warehouse has not been provisioned and Docker daemon is unavailable

API and Format

No API or storage format changes.

Documentation

No documentation changes required.

[QuakeWang]

Nit: this helper compares the full pretty_format_batches output verbatim. It keeps the expected SQL result readable, but also couples the tests to Arrow/DataFusion pretty-printer details such as padding, borders, and NULL rendering. Fine for these small result sets; if we want less brittle tests across dependency upgrades, comparing normalized rows/columns would be more stable.