GH-49785: [C++][FlightRPC][ODBC] Get ODBC tests passing on Linux

[justing-bq]

Rationale for this change

After getting the tests building on Linux, the next step is to get all the tests passing.
Resolves #49785.

What changes are included in this PR?

Miscellaneous fixes.

Are these changes tested?

Yes.

Are there any user-facing changes?

No.

Depends on #49668 and #49784 before being ready for review.

• GitHub Issue: [C++][FlightRPC][ODBC] Get ODBC tests passing on Linux #49785

[github-actions]

⚠️ GitHub issue #49785 has been automatically assigned in GitHub to PR creator.

[justing-bq]

@lidavidm @kou
Please take a look when you have time.

[kou]

Is the ODBC Linux job failure tracked?
https://github.com/apache/arrow/actions/runs/26246962168/job/77247867378?pr=49786

[Copilot]

Pull request overview

This PR adjusts the Flight SQL ODBC C++ test suite to pass on Linux by updating platform-specific expectations, selectively disabling known-problematic tests on Linux, and re-enabling the ODBC test target in Linux CI.

Changes:

• Update multiple ODBC tests to reflect Linux/Unix-ODBC vs Windows vs macOS driver-manager behavior (return codes, SQLSTATEs, and reported info strings).
• Disable a subset of metadata-related tests on Linux due to an acknowledged BlockingQueue issue (GH-49702).
• Re-enable arrow-flight-sql-odbc-test execution on Linux in CI.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
cpp/src/arrow/flight/sql/odbc/tests/type_info_test.cc	Adjust expected SQLGetTypeInfo results for SQL_TYPE_* across Windows vs non-Windows.
cpp/src/arrow/flight/sql/odbc/tests/tables_test.cc	Reorder typed tests and disable several table-metadata tests on Linux (GH-49702).
cpp/src/arrow/flight/sql/odbc/tests/statement_test.cc	Adjust varbinary byte comparisons and add Linux-specific expectations for some APIs.
cpp/src/arrow/flight/sql/odbc/tests/statement_attr_test.cc	Linux-specific SQLSTATE expectation for an unsupported stmt attribute.
cpp/src/arrow/flight/sql/odbc/tests/odbc_test_suite.h	Add kErrorState00000 constant for tests expecting a “success” SQLSTATE.
cpp/src/arrow/flight/sql/odbc/tests/get_functions_test.cc	Restrict SQLGetFunctions tests to Windows due to Unix driver-manager limitations.
cpp/src/arrow/flight/sql/odbc/tests/errors_test.cc	Platform-specific expectations for invalid-handle behavior and SQLSTATE differences.
cpp/src/arrow/flight/sql/odbc/tests/connection_info_test.cc	Disable one info test on Linux and adjust expected SQL_ODBC_VER values.
cpp/src/arrow/flight/sql/odbc/tests/connection_attr_test.cc	Add pre-connect fixture and adjust platform-specific behavior for several connection attributes.
cpp/src/arrow/flight/sql/odbc/tests/columns_test.cc	Disable column-metadata helpers/tests on Linux (GH-49702) and tighten Unicode column checks.
ci/scripts/cpp_test.sh	Stop excluding arrow-flight-sql-odbc-test on Linux.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[justing-bq]

Is the ODBC Linux job failure tracked? https://github.com/apache/arrow/actions/runs/26246962168/job/77247867378?pr=49786

This failure is due to a problem with arrow-flight-test.
arrow-flight-sql-odbc-test is passing fine.

[alinaliBQ]

Is the ODBC Linux job failure tracked? https://github.com/apache/arrow/actions/runs/26246962168/job/77247867378?pr=49786

This failure is due to a problem with arrow-flight-test. arrow-flight-sql-odbc-test is passing fine.

Yup, and the failure is tracked in issue #49767. cc @kou

[justing-bq]

@kou @lidavidm
Comments addressed. Please review.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 3 comments.

[justing-bq]

@kou @lidavidm
Comments addressed. Please review.

[lidavidm]

Looks like the test failure is in Gandiva and is unrelated.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 16fe342.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 6 possible false positives for unstable benchmarks that are known to sometimes produce them.