added support for MapFromEntries

[athlcode]

Which issue does this PR close?

Closes # (none: prerequisite for apache/datafusion-comet#2706; follow-up to #17779 and #19274).

Rationale for this change

The existing Spark map_from_entries / map_from_arrays UDFs silently kept the last occurrence of a duplicate key, hardcoded to the LAST_WIN branch of Spark's spark.sql.mapKeyDedupPolicy. Spark's default is EXCEPTION, and Spark 4 raises SparkRuntimeException with error class DUPLICATED_MAP_KEY. Without matching the default, downstream engines (e.g. datafusion-comet) have to fall back to Spark even for the common case.

What changes are included in this PR?

1. In map_deduplicate_keys (utils.rs), raise [DUPLICATED_MAP_KEY] Duplicate map key {key} was found ... when a duplicate is encountered, matching Spark's default behavior and error class. The LAST_WIN branch is removed along with its TODO comment.
2. Update the two affected sqllogictest files to assert the new DUPLICATED_MAP_KEY error instead of the previous last-wins output.

No config-option or enum is introduced at this stage. map_from_entries.rs, map_from_arrays.rs, and config.rs are untouched.

Are these changes tested?

Yes. The duplicate-key assertions in sqllogictest/test_files/spark/map/map_from_entries.slt and sqllogictest/test_files/spark/map/map_from_arrays.slt are flipped from positive-result to query error.

Are there any user-facing changes?

Yes, a behavior change: duplicate keys now raise [DUPLICATED_MAP_KEY] under the default policy instead of silently collapsing to the last occurrence. This aligns with Spark's documented default. No new config keys, no API changes.