fix!: Add support for Athena's Iceberg partitioning transforms

[VaggelisD]

Follow up on #4962

This PR introduces the expressions exp.PartitionedByBucket and exp.PartitionedByTruncate which control the generation of Iceberg transforms:

• Athena's "Hive" engine (used for CREATE TABLE) demands BUCKET(n, col) / TRUNCATE(n, col)
• Athena's "Trino" engine (used for CTAS) demands BUCKET(col, n) / TRUNCATE(col, n)

Note: The PARTITIONED BY property attempts to parse a schema, so the new expressions are parsed as "constraints" since _parse_schema() ends up parsing _parse_constraint() or _parse_column_def().

[georgesittas]

Let's verify that this fixes the linked SQLMesh issue.

[VaggelisD]

Let's verify that this fixes the linked SQLMesh issue.

If I'm not mistaken this is one part of the fix; The other part would be parsing the partition_by model expression into the new nodes, which will be generated into the proper Athena engine by using the appropriate generator.

[georgesittas]

I see, makes sense, thanks for clarifying.

[erindru]

This looks correct, thanks!