GH-34535: [C++] Move ChunkResolver to the public API

[anjakefala]

Rationale for this change

Adopting #40226.

The creation and return of a shared_ptr does result in some performance overhead, that makes a difference for a performance-sensitive application.

If someone could use ChunkResolver to learn the indices, they could then instead access the data directly.

What changes are included in this PR?

• Updates to documentation (thanks to @SChakravorti21 )
• Moving ChunkResolver to public API, and updating all references to it in the code

Are these changes tested?

There seemed to be comprehensive tests already: https://github.com/apache/arrow/blob/main/cpp/src/arrow/chunked_array_test.cc#L324 If an edgecase is missing, I'd be happy to add it.

Are there any user-facing changes?

ChunkResolver and TypedChunkLocation are now in the public API.

• GitHub Issue: [C++] Request to move ChunkResolver to public API #34535

[anjakefala]

I'm aware of the build failures and actively resolving them!

[anjakefala]

For the failing R builds, this conversation needs to be resolved: #43623

[felipecrv]

Yes, I agree this API should be made public. I've added tests and benchmarks recently [1].

[1] #43954

[anjakefala]

Please add a Co-authored-by: SChakravorti21 <[email protected]> in the squashed commit. I tried to add it, but I seem to had a typo, and it didn't register.

(Edit: Turns out PRs are merged via a script, and not a manual squash.)

[anjakefala]

I amended @SChakravorti21 as an author in a recent commit, so that the merge script will pick them up as a co-author.

[anjakefala]

@assignUser The Java failure seems to be a network error! Is there a chance of merging this for the release?

[assignUser]

@anjakefala I'll re-run the job. This was not marked as a blocker and with rc0 already cut I will defer to @raulcd if we can add this to rc1 (if it will happen).

[assignUser]

There seem to be some actual java test errors now, I don't think they are related but am not ☕ to judge that.
https://github.com/apache/arrow/actions/runs/11445780195/job/31852155160?pr=44357#step:7:10025

[felipecrv]

Is anyone against me merging this one?

[felipecrv]

@anjakefala fix version was set to 19.0.0. You might want to change it on the issue to 18.0.0.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 8208774.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 15 possible false positives for unstable benchmarks that are known to sometimes produce them.