[SPARK-55356][SQL] Support alias for PIVOT clause

[filipdavidovic]

What changes were proposed in this pull request?

Add support for alias in PIVOT clause, following the same syntax as T-SQL and BigQuery.

Short spec to describe the improvement:

Extend the PIVOT clause to accept an optional alias (PIVOT (...) AS alias), allowing the pivoted result set to be referenced by name in the surrounding query context. The alias qualifies all columns produced by the pivot operation, including both the grouping columns and the dynamically generated value columns. When an alias is provided, columns may be referenced as alias.column_name. When omitted, current unqualified resolution behaviour is preserved. In queries with multiple pivots or joins, the alias disambiguates column references. The alias scope is limited to the query block in which the pivot appears.

Why are the changes needed?

Main reason for the improvement is parity with platforms like SQL Server. When customers want to migrate, their queries using PIVOT alias can't be automatically transpiled, requiring a human in the loop.

Does this PR introduce any user-facing change?

Yes, it introduces an optional alias and updates the documentation to reflect it. Change is fully backwards compatible.

How was this patch tested?

Added end-to-end tests in SQLQueryTestSuite and unit tests in PivotParserSuite.

build/sbt "catalyst/testOnly *PivotParserSuite"

build/sbt "sql/testOnly *SQLQueryTestSuite -- -z pivot.sql"

Was this patch authored or co-authored using generative AI tooling?

Implemented with help of Claude Code.

[github-actions]

JIRA Issue Information

=== Improvement SPARK-55356 ===
Summary: Support alias in Spark SQL PIVOT
Assignee: None
Status: Open
Affected: ["4.1.1"]

---

This comment was automatically generated by GitHub Actions

[cloud-fan]

nit: SQL golden file is for integration test. Tiny cases like this should go to unit test (which we already did and we can just remove this query here)

[filipdavidovic]

Moved most of the tests from golden file to unit test suite. Left two complex ones for sanity checks.