GH-49998: [CI][Python] Pin to an older release of miniforge to fix mamba hang

[adamreeve]

Rationale for this change

See #49998. This prevents CI jobs from timing out after an hour.

What changes are included in this PR?

Pins miniforge to 26.1.1-3.

Are these changes tested?

Yes, I tested locally that this fixes the slow install.

Are there any user-facing changes?

No, this only affects CI.

• GitHub Issue: [CI][Python] AMD64 Conda Python 3.10 Pandas 1.3.4 job consistently timing out #49998

[github-actions]

⚠️ GitHub issue #49998 has been automatically assigned in GitHub to PR creator.

[raulcd]

The current CI failures is unrelated:

• #50002

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 2984d23.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details.

[adamreeve]

Thanks @raulcd. I've made #50004 to follow up and revert this later once the mamba issue is fixed.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 2984d23.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 2 possible false positives for unstable benchmarks that are known to sometimes produce them.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 2984d23.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 2 possible false positives for unstable benchmarks that are known to sometimes produce them.