Leverage Iceberg-Rust for all the transforms

[Fokko]

Rationale for this change

Testing out to use Iceberg Rust for all of the transforms. I think we have some rounding error in apache/iceberg-rust#1128

Closes #1591

Are these changes tested?

Are there any user-facing changes?

[kevinjqliu]

LGTM! I fixed CI and added pyiceberg-core to the pyarrow install group for poetry

[tool.poetry.extras]
pyarrow = ["pyarrow", "pyiceberg-core"]

[Fokko]

technically this is not required, because we only use the transforms when writing to a partitioned table. But I think that it might lead to a lot of confusion if we don't do this.

[kevinjqliu]

Agreed i was considering this too and came to the same conclusion :)

Also I think the pyarrow_transform functions are used on the read path too

iceberg-python/pyiceberg/io/pyarrow.py

Line 2696 in a67c559

name, partition.transform.pyarrow_transform(source_field.field_type)(arrow_table[source_field.name])

[Fokko]

Also I think the pyarrow_transform functions are used on the read path too

For completeness, I don't think that's true. _determine_partitions is only used by _dataframe_to_data_files. When reading, we often have a single value SELECT * FROM tbl WHERE created_at > '2025-01-01 19:25:00', so there performance is not that important, and we use non-Arrow transform (example for the MonthPartitioning).