[Python][Docs][CI] Disabled numpydoc PR02 check may skip PR03

### Describe the bug, including details regarding any error messages, version, and platform.

numpydoc validation only runs parameter order PR03 check when docstring and signature parameter match - when PR02 Unknown parameter is hit currently that doesn't get reported but more importantly PR03 is skipped in that case (e.g. https://github.com/apache/arrow/pull/47498#issuecomment-5601019229)

https://github.com/apache/arrow/blob/0d8f1b9929462364b4fe48a68891d468e0b156a8/compose.yaml#L1559
https://numpydoc.readthedocs.io/en/latest/validation.html

After [enabling PR02 on my fork](https://github.com/tadeja/arrow/actions/runs/34332116415/job/102402892816#step:6:4425) :
```shell

pyarrow._dataset_parquet_encryption.ParquetEncryptionConfig
PR02: Unknown parameters {'encryption_config', 'kms_connection_config', 'crypto_factory'}

pyarrow._dataset_parquet_encryption.ParquetDecryptionConfig
PR02: Unknown parameters {'decryption_config', 'kms_connection_config', 'crypto_factory'}

pyarrow.lib.OSFile
PR02: Unknown parameters {'path'}

pyarrow.parquet.core.write_table
PR02: Unknown parameters {'use_content_defined_chunking'}

pyarrow.parquet.core.read_pandas
PR02: Unknown parameters {'schema', 'thrift_container_size_limit', 'page_checksum_verification', 'arrow_extensions_enabled', 'decryption_properties', 'coerce_int96_timestamp_unit', 'buffer_size', 'filesystem', 'filters', 'list_type', 'partitioning', 'thrift_string_size_limit', 'read_dictionary', 'memory_map', 'pre_buffer', 'use_threads', 'ignore_prefixes', 'binary_type'}

pyarrow.parquet.core.ParquetWriter
PR02: Unknown parameters {'data_page_size', 'use_content_defined_chunking', 'bloom_filter_options', 'coerce_timestamps', 'allow_truncated_timestamps'}

pyarrow._s3fs.S3FileSystem
-> pyarrow._s3fs.S3FileSystem(access_key=None, *, secret_key=None, session_token=None, bool anonymous=False, region=None, request_timeout=None, connect_timeout=None, scheme=None, endpoint_override=None, bool background_writes=True, default_metadata=None, role_arn=None, session_name=None, external_id=None, load_frequency=900, proxy_options=None, allow_delayed_open=False, allow_bucket_creation=False, allow_bucket_deletion=False, check_directory_existence_before_creation=False, retry_strategy: S3RetryStrategy = AwsStandardS3RetryStrategy(max_attempts=3), force_virtual_addressing=False, tls_ca_file_path=None)
PR02: Unknown parameters {'background_writes', 'check_directory_existence_before_creation', 'external_id', 'force_virtual_addressing', 'allow_bucket_creation', 'retry_strategy', 'region', 'connect_timeout', 'access_key', 'scheme', 'load_frequency', 'endpoint_override', 'session_name', 'request_timeout', 'allow_delayed_open', 'tls_ca_file_path', 'allow_bucket_deletion', 'role_arn', 'anonymous', 'proxy_options', 'secret_key', 'default_metadata', 'session_token'}

pyarrow._fs.LocalFileSystem
-> pyarrow._fs.LocalFileSystem(use_mmap=False, *)
PR02: Unknown parameters {'use_mmap'}

Total number of docstring violations: 8

```

Great summary in open issue #33646
but PR02 is not listed there as planned, so PR02 skipping PR03 can be reviewed here.

### Component(s)

Python, Documentation, Continuous Integration