GH-46373: [Python] Exercise fallback case on tests for parquet.read_table in case dataset is not available

[raulcd]

Rationale for this change

Since we merged:

• GH-31303: [Python] Remove the legacy ParquetDataset custom python-based implementation #39112

The fallback case were dataset is not available is not being tested on parquet.read_table (I did validate myself with a breakpoint).

What changes are included in this PR?

Adds a basic test that checks the exercises that code path and tests the read_table method from ParquetFile instead of ParquetDataset in case dataset is not available.

Are these changes tested?

Yes via CI.

Are there any user-facing changes?

No

• GitHub Issue: [Python] Exercise fallback case on tests for parquet.read_table in case dataset is not available #46373

[github-actions]

⚠️ GitHub issue #46373 has been automatically assigned in GitHub to PR creator.

[AlenkaF]

Looks great, thanks for adding this @raulcd!

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit d028fa8.

There weren't enough matching historic benchmark results to make a call on whether there were regressions.

The full Conbench report has more details.