https://issues.apache.org/jira/browse/ARROW-14306

⚠️ Ticket has not been started in JIRA, please click 'Start Progress'.

Oh, and it needs to be listed in https://github.com/apache/arrow/blob/master/docs/source/python/api/compute.rst as well (again alongside its ascii/utf8 counterparts)

Thanks for the comments. I added the docs. I also added a test because the function can be applied to String/LargeString as well. I'm in doubt whether we would want to prevent applying this function to String/LargeString. Use cases would be quite exotic.

If we'd like to prevent it I could introduce BinaryTypes() with {binary(), large_binary()} in arrow/type.h and use that to add the kernel to the registry only for these types, but I'm not sure if this would have further implications for the rest of the codebase.

Oh, whoops. No, good point: we should not support string types here, we should register only Binary and LargeBinary. For String/LargeString the kernel would need to validate that its output is still valid UTF-8 which would be redundant with string_reverse.

Benchmark runs are scheduled for baseline = 00d5077 and contender = b1f009c. b1f009c is a master commit associated with this PR. Results will be available as each benchmark for each run completes.
Conbench compare runs links:
[Finished ⬇️0.0% ⬆️0.0%] ec2-t3-xlarge-us-east-2
[Failed ⬇️0.0% ⬆️0.0%] ursa-i9-9960x
[Finished ⬇️0.09% ⬆️0.0%] ursa-thinkcentre-m75q
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python. Runs only benchmarks with cloud = True
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java