fix flaky test by adding retry logic to fetch openml dataset in notebook

[imatiach-msft]

Description

One of my PRs failed on a flaky test validating the notebook fairness-dashboard-loan-allocation.ipynb:

https://github.com/microsoft/responsible-ai-toolbox/runs/7284751274?check_suite_focus=true#step:11:266

The error was a request failure:
URLError: <urlopen error [Errno 104] Connection reset by peer>
When calling fetch_openml function.

This PR resolves the transient failures by adding retry logic, which should make the notebook test fail much less often.

Checklist

• I have added screenshots above for all UI changes.
• I have added e2e tests for all UI changes.
• Documentation was updated if it was needed.

[codecov-commenter]

Codecov Report

Merging #1547 (1065b9c) into main (1a435ca) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main    #1547   +/-   ##
=======================================
  Coverage   87.39%   87.39%           
=======================================
  Files         107      107           
  Lines        5069     5069           
=======================================
  Hits         4430     4430           
  Misses        639      639           

Flag	Coverage Δ
unittests	87.39% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1a435ca...1065b9c. Read the comment docs.

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-openml-fetch-err/dashboard/index.html