GH-35599: [Python] Canonical fixed-shape tensor extension array/type is not picklable.

[AlenkaF]

This PR adds __reduce__ method to the FixedShapeTensorType.

• Closes: [Python] Canonical fixed-shape tensor extension array/type is not picklable. #35599

[jorisvandenbossche]

Thanks!

[ursabot]

Benchmark runs are scheduled for baseline = 98559fe and contender = 059d8b6. 059d8b6 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
[Finished ⬇️0.62% ⬆️0.09%] test-mac-arm
[Finished ⬇️0.33% ⬆️0.0%] ursa-i9-9960x
[Finished ⬇️0.39% ⬆️0.03%] ursa-thinkcentre-m75q
Buildkite builds:
[Finished] 059d8b66 ec2-t3-xlarge-us-east-2
[Finished] 059d8b66 test-mac-arm
[Finished] 059d8b66 ursa-i9-9960x
[Finished] 059d8b66 ursa-thinkcentre-m75q
[Finished] 98559fe0 ec2-t3-xlarge-us-east-2
[Finished] 98559fe0 test-mac-arm
[Finished] 98559fe0 ursa-i9-9960x
[Finished] 98559fe0 ursa-thinkcentre-m75q
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python, R. Runs only benchmarks with cloud = True
test-mac-arm: Supported benchmark langs: C++, Python, R
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java

[pitrou]

Instead of defining __reduce__ for each concrete extension type, why not define it on BaseExtensionType by making of use of C++-level (de)serialization?

[jorisvandenbossche]

We actually did that somewhat as a follow-up in #36170, defining __reduce__ on ExtensionType (which can be moved to BaseExtensionType), but with using the Python APIs instead of the C++ APIs (so using self.__arrow_ext_deserialize__ instead of self.ext_type.Deserialize), which then requires those dunders to be implemented (even when it is wrapping one defined in C++). This can indeed be cleaned up.

[pitrou]

A cleanup would nice indeed, to avoid such spurious additions.

[AlenkaF]

Opened a new issue: #39094