[python] Fix the row filter on PyArrow 6

[sundapeng]

Purpose

#10539 made FilterRecordBatchReader._filter_batch_by_row call RecordBatch.select. PyArrow 6, which the Python 3.6 job installs, has Table.select but not RecordBatch.select, so that job fails:

FAILED pypaimon/tests/py36/rest_ao_read_write_test.py::RESTAOReadWritePy36Test::test_append_only_reader_with_filter
E       AttributeError: 'pyarrow.lib.RecordBatch' object has no attribute 'select'
pypaimon/read/reader/filter_record_batch_reader.py:100: AttributeError

For example, job 115681294594 of #10552. Master's Python jobs stopped at flake8 before #10552 fixed it, so this failure only showed up afterwards.

Master now fails the same way, for example job 116061121154 on b634373. #10572 contains the same change, but it is waiting for apache/paimon-rust#1128.

This pull request selects the columns on the Table that the next line builds anyway.

Tests

• CI's Python 3.6 selection (Python 3.6, PyArrow 6.0.1): 1 failed and 119 passed on master, 120 passed with this change (rechecked on b634373).
• Python 3.11, the 43 test files that cover the row filter and query auth: the same results as on master.
• flake8 with dev/cfg.ini passes.

API and Format

No.

Documentation

No.

[QuakeWang]

+1