test: add schema evolution time travel coverage

[liujiwen-up]

Purpose

Linked issue: #258

Add stable time-travel coverage for schema evolution boundaries using tag-based versions, without relying on warehouse-global snapshot numbers.

Brief change log

• Add a Spark fixture for time_travel_schema_evolution with tags around add, rename, drop, and reorder schema changes.
• Add core integration coverage using scan.version tag reads with the time-travel schema resolution path.
• Add DataFusion SQL coverage using VERSION AS OF 'tag_name' with SELECT, WHERE, and ORDER BY.
• Make Spark warehouse cleanup retry briefly to avoid Docker/overlay directory deletion races during reprovisioning.

Tests

• python3 -m py_compile dev/spark/provision.py: passed
• cargo fmt --check: passed
• cargo test -p paimon-integration-tests --test read_tables time_travel_schema_evolution --no-run: passed
• cargo test -p paimon-datafusion --test read_tables time_travel_schema_evolution --no-run: passed
• make docker-up: passed
• cargo test -p paimon-integration-tests --test read_tables time_travel_schema_evolution -- --nocapture: passed
• cargo test -p paimon-datafusion --test read_tables time_travel_schema_evolution -- --nocapture: passed
• git diff --check: passed

API and Format

No API or storage format changes.

Documentation

No documentation update required; this is test fixture and coverage only.

[QuakeWang]

+1