[Python] pyarrow.compute functions don't convert args with __arrow_array__.

When the compute functions pack args, lists and numpy arrays are converted into arrow arrays, but objects with `__arrow_array__` defined aren't. A TypeError is raised.

**Reporter**: [A. Coady](https://issues.apache.org/jira/browse/ARROW-15210) / @coady
#### Related issues:
- [[Python] Create pyarrow array using an object's `__array__` method.](https://github.com/apache/arrow/issues/30703) (is related to)
#### PRs and other links:
- [GitHub Pull Request #12048](https://github.com/apache/arrow/pull/12048)

_{**Note**: *This issue was originally created as [ARROW-15210](https://issues.apache.org/jira/browse/ARROW-15210). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*}

Start by tracing how pyarrow.compute functions pack arguments and compare conversion of lists and NumPy arrays with objects defining __arrow_array__. Reproduce the TypeError, then verify that those objects are converted successfully by the compute functions without regressing existing argument handling.

Written by the indexing model from the issue text.