Dictionary-encoded Timestamp / TimestampNTZ not handled in extract_date_part #3324
Description
Describe the problem
While working on #3278, we noticed that extract_date_part only matches
DataType::Timestamp(_, None) and DataType::Timestamp(_, Some(_)).
Dictionary-encoded timestamps (e.g. DataType::Dictionary(_, Timestamp))
would currently fall into the error branch.
This may affect Iceberg tables where timestamp columns are often dictionary-encoded.
Expected behavior
extract_date_part should correctly handle dictionary-backed Timestamp and TimestampNTZ inputs,
similar to other Spark expressions.
Context
This was discussed during review of #3278 and is intentionally tracked separately
to keep that PR focused.
Proposed follow-up work
- Add dictionary-aware handling in
extract_date_part - Add a regression test (likely Spark-side) covering dictionary-encoded TimestampNTZ