GH-36187: [C++] Display the name of the problematic field when returning status "Data type ... is not supported in join non-key field" for HashJoin

[vibhatha]

Rationale for this change

The change is bases on a user filed issue. This would provide a readable error message.

What changes are included in this PR?

Did modification to hash_join_node.cc to include field name to the already provided error message.
Added python tests to validate the response.

Are these changes tested?

A test case has been added to Python under test_table.py

Are there any user-facing changes?

No, just a clear error message is provided.

• Closes: [C++] Display the name of the problematic field when returning status "Data type ... is not supported in join non-key field" for HashJoin #36187

[westonpace]

This looks good to me modulo a few very minor nits.

[kou]

We need to add @pytest.mark.acero because this depends on Acero.

I'm adding it in #36681 but we may want to do it in a separated PR only for it...

[kou]

I've split the fix to #36683.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 6 benchmarking runs that have been run so far on merge-commit 95e3e77.

There were 3 benchmark results indicating a performance regression:

• Commit Run on ursa-i9-9960x at 2023-07-22 01:05:53Z
  • engine=arrow, format=parquet, language=R, memory_map=False, query_id=TPCH-01, scale_factor=1
  • engine=arrow, format=parquet, language=R, memory_map=False, query_id=TPCH-03, scale_factor=1
• and 1 more (see the report linked below)

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