test: cover mixed-format reorder column reads

[liujiwen-up]

Purpose

Linked issue: #258

Add focused Spark-provisioned integration coverage for mixed-format schema evolution reads after ALTER COLUMN ... FIRST/AFTER.

This PR intentionally covers only the reorder/move-column case to avoid overlapping with #354, which covers mixed-format add-column and type-promotion reads.

Brief change log

• Add a Spark fixture with Parquet, ORC, and Avro files written across column reorder/move schema versions.
• Add an integration test verifying full reads expose the current table schema order.
• Verify mixed-format reads map values by field id rather than physical column position.
• Verify projected reads preserve caller-specified projection order after column reorder/move.

Tests

• cargo fmt: passed
• git diff --check: passed
• cargo test -p paimon-integration-tests --test read_tables --no-run: passed

API and Format

No API or storage format changes. This PR only adds Spark provisioning and Rust integration test coverage.

Documentation

No documentation update required.