Python: Add PartitionField

[dramaticlly]

This is 1st step of reintroduce #3228

adds PartitionField in iceberg/partitioning.py

>>> from iceberg.partitioning import PartitionField
>>> partition_field = PartitionField(3, 1, None, "id")
>>> partition_field
PartitionField(field_id=1, name=id, transform=None, source_id=3)

Purposefully keep the change list small to focus only on PartitionField and once checked in we can look into PartitionSpec next.

CC @samredai @rdblue @jun-he

[samredai]

Overall this looks good, I suggested a few changes. Also I'm wondering if this should go into the table sub-module, I was thinking table/partition.py.

[samredai]

@rdblue do we need to define a hash method for this class? Maybe something like:

def __hash__(self):
    return hash((self.source_id, self.field_id, self.name, self.transform))

[dramaticlly]

Looks like it's mainly used in list or object array from what I can tell, but please let me know if I should add it!

[szehon-ho]

Not sure the policy for adding hash across python classes, but maybe it will be useful as PartitionField is part of Iceberg public api ,(example if a user tries to make dicts keyed by partitionSpec?) Java version has it, for comparison.

[rdblue]

Sorry I was late to see this, but we don't need __hash__ implementations unless the objects are being used for keys in dicts.

[dramaticlly]

@samredai added hash and moved partitioning to the table module, if you could take another look :)

[dramaticlly]

@rdblue and @samredai updated to address review feedback, appreciate another look :)

[rdblue]

Looks great. Thanks, @dramaticlly!