[lake/iceberg] Support custom Iceberg lake table path

[litiliu]

Purpose

Linked issue: close #4251

Fluss currently requires an Iceberg lake table to use the same database and table name as the Fluss table. This change extends the existing table.datalake.database-name and table.datalake.table-name options to Iceberg, allowing the physical Iceberg table path to differ from the logical Fluss table path.

Paimon already supports configuring a custom physical lake table path through #3523. This change provides the corresponding capability for Iceberg.

Brief change log

• Allow custom lake database and table names for Iceberg tables.
• Resolve the mapped Iceberg table path when validating, writing, and committing tiered data.
• Prevent an existing Iceberg table from being rebound to a different physical table path.
• Preserve the resolved physical path for Flink $lake tables, including Iceberg metadata tables such as $lake$snapshots and union reads.
• Keep the existing same-name behavior when the mapping options are not configured.

Tests

• Added validation coverage for supported and unsupported lake formats.
• Added Iceberg catalog tests for mapping changes before and after the physical table is created.
• Added an Iceberg tiering test covering validator, writer, and committer access through a mapped physical table path.
• Added an Iceberg Flink end-to-end test covering tiering, the physical table mapping, $lake, $lake$snapshots, and streaming union read. Its asynchronous result collection and job cancellation have explicit one-minute deadlines.
• Ran the targeted Flink catalog unit tests (22 tests passed).
• Ran FlinkUnionReadPrimaryKeyTableITCase#testUnionReadWithCustomLakeTablePath (passed).
• Ran TableDescriptorValidationTest, IcebergLakeCatalogTest, and IcebergTieringTest (73 tests passed).

API and Format

No public Java API or storage format changes. The existing table-level lake path options now also support Iceberg.

Documentation

Updated the configuration descriptions for table.datalake.database-name and table.datalake.table-name to include Iceberg.

Generative AI disclosure

• Yes. OpenAI Codex was used to assist with implementation, review, and testing.

[litiliu]

@luoyuxia Please take a look when you have time. Thanks!