GH-45816: [C++] Make VisitType() fallback branch unreachable

[kszucs]

Rationale for this change

Enable using the more convenient constexpr type checking functions from type_traits with VisitType dispatchers, like:

  auto handle_type = [&](auto&& type) {
    using Type = std::decay_t<decltype(type)>;
    if constexpr (::arrow::is_boolean(Type::type_id)) {
      ...
    }
    else if constexpr (::arrow::is_primitive(Type::type_id)) {
        ...
    }
    /* else etc. */
  };
  return VisitType(*values.type(), handle_type);

What changes are included in this PR?

Remove the fallback call with const DataType& rather make the default branch unreachable at compile time.

Are these changes tested?

Yes, VisitType is being used at multiple places in the codebase.

Are there any user-facing changes?

Yes, this is a breaking change. The visitor passed to VisitType doesn't need to have an implementation for the base DataType.

• GitHub Issue: [C++] VisitType() requires the visitor to have in implementation for base DataType #45816

[github-actions]

⚠️ GitHub issue #45816 has been automatically assigned in GitHub to PR creator.

[pitrou]

I think this is a good idea, let's see if it breaks CI.

[pitrou]

@github-actions crossbow submit -g cpp

[github-actions]

Revision: 0677196

Submitted crossbow builds: ursacomputing/crossbow @ actions-7723444ace

Task	Status
example-cpp-minimal-build-static
example-cpp-minimal-build-static-system-dependency
example-cpp-tutorial
test-alpine-linux-cpp
test-build-cpp-fuzz
test-conda-cpp
test-conda-cpp-meson
test-conda-cpp-valgrind
test-cuda-cpp-ubuntu-22.04-cuda-11.7.1
test-debian-12-cpp-amd64
test-debian-12-cpp-i386
test-fedora-39-cpp
test-ubuntu-22.04-cpp
test-ubuntu-22.04-cpp-20
test-ubuntu-22.04-cpp-bundled
test-ubuntu-22.04-cpp-emscripten
test-ubuntu-22.04-cpp-no-threading
test-ubuntu-24.04-cpp
test-ubuntu-24.04-cpp-bundled-offline
test-ubuntu-24.04-cpp-gcc-13-bundled
test-ubuntu-24.04-cpp-gcc-14
test-ubuntu-24.04-cpp-minimal-with-formats
test-ubuntu-24.04-cpp-thread-sanitizer

[kszucs]

Shall we explicitly test the following use case?

auto handle_type = [&](auto&& type) {
    using Type = std::decay_t<decltype(type)>;
    if constexpr (::arrow::is_boolean(Type::type_id)) {
      ...
    }
    else if constexpr (::arrow::is_primitive(Type::type_id)) {
        ...
    }
    /* else etc. */
  };
  return VisitType(*values.type(), handle_type);

If yes, then where? Perhaps modify an existing use case, like using type_id based type checkers in arrow::csv::QuotedColumnPopulator::MakePopulator?

[pitrou]

IMHO we don't have to.