[C++][Python] The ORC writer does not support all-null object columns from Pandas

Object columns in Pandas are currently converted to Null type columns in Arrow and they can not be written using `pyarrow.orc.write_table`. Since ORC doesn't support Null type columns we need to find the appropriate DataType for such columns and do the write.

**Reporter**: [Ian Alexander Joiner](https://issues.apache.org/jira/browse/ARROW-14782) / @iajoiner

_{**Note**: *This issue was originally created as [ARROW-14782](https://issues.apache.org/jira/browse/ARROW-14782). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*}

Start at pyarrow.orc.write_table and trace how Pandas object columns become Arrow Null type columns before reaching the ORC writer. Determine an appropriate DataType for all-null object columns and verify that such columns can be written successfully without the ORC Null type limitation.

Written by the indexing model from the issue text.