[python] Support null blob

[XiaoHongbo-Hope]

Purpose

Tests

[leaves12138]

+1

[leaves12138]

Thanks for the patch. The null encoding/decoding is aligned with the Java blob format.

One edge case: BlobFormatWriter.write_value accepts col_data is None before validating that the field is actually a BLOB field. This means a non-BLOB single-column Arrow table with all-null values, e.g. pa.int64() or pa.string(), can now be written as a valid blob file, while non-null values of the same type are still rejected. Could we validate the field type before accepting nulls, or add a write_blob-level check that only large_binary is supported?

Also, the file_io.write_blob null cases currently only assert that the file exists. It would be better to add a public API round-trip test, e.g. write [b"data", None, b"more"] through write_blob and read it back with FormatBlobReader to verify the null value is preserved.