GH-44795: [C++] Use arrow::util::span on arrow::util::bitmap_builders_utilities instead of std::vector

[raulcd]

Rationale for this change

arrow::util::span (a backport of C++20 std::span) is more generally applicable than std::vector, so any public API currently accepting a vector const-ref argument should instead accept a span argument.

What changes are included in this PR?

arrow::util::BytesToBits accepts arrow::util::span instead of std::vector

Are these changes tested?

Yes, existing C++ tests via CI

Are there any user-facing changes?

Yes, from Result<std::shared_ptr<Buffer>> BytesToBits(const std::vector<uint8_t>&, MemoryPool* pool) to Result<std::shared_ptr<Buffer>> BytesToBits(util::span<const uint8_t> bytes, MemoryPool* pool)

• GitHub Issue: [C++] APIs taking a const vector<T>& argument should instead take a span<const T> #44795

[github-actions]

⚠️ GitHub issue #44795 has been automatically assigned in GitHub to PR creator.

[raulcd]

The macOS CI failures are unrelated, I opened: #44797
@pitrou is something like this what you had in mind? I did it for a single API method just to validate I understood what you had in mind. I can open subtasks or aggregate some more on a single PR.

[pitrou]

@raulcd Yes, this is the kind of change I had in mind. See comments.

[pitrou]

LGTM, thank you @raulcd

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 7af2ad4.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 3 possible false positives for unstable benchmarks that are known to sometimes produce them.