feature: Add dataset fetch retry and update tests with new word sets references

[pbadillatorrealba]

This pull request updates several queries and tests to replace the unpleasant_5 word set with a new unpleasant_5a word set. Additionally, it introduces a new _retry_request function with comprehensive test coverage for retrying network requests. Below is a breakdown of the most important changes grouped by theme:

Query Updates:

• Updated multiple queries in docs/user_guide/measurement_user_guide.rst and docs/user_guide/mitigation_user_guide.rst to use unpleasant_5a instead of unpleasant_5 for bias measurements. [1] [2] [3] [4]

Test Updates:

• Updated test fixtures in tests/debias/conftest.py and tests/metrics/conftest.py to reflect the change from unpleasant_5 to unpleasant_5a in various queries. [1] [2] [3] [4]
• Updated test cases in tests/debias and tests/test_preprocessing.py to ensure compatibility with the new unpleasant_5a word set. [1] [2] [3] [4]

New Retry Functionality:

• Added _retry_request function in tests/test_datasets.py to handle retries for network requests, including support for exponential backoff and handling specific exceptions like HTTP 429 (rate limit) and timeout errors.
• Added a comprehensive suite of tests for _retry_request to validate its behavior under various scenarios, such as success on first attempt, rate limit errors, timeout errors, and non-retryable HTTP errors.

Dataset Updates:

• Updated dataset-related tests in tests/test_datasets.py to include the new unpleasant_5a and unpleasant_5b word sets for validation.

These changes ensure consistency across the codebase for the updated word set and enhance the robustness of network request handling.

[Copilot]

Pull Request Overview

This PR updates word set references from unpleasant_5 to unpleasant_5a throughout the codebase to maintain consistency with updated word sets.

Key Changes:

• Replaced all references to unpleasant_5 with unpleasant_5a across test files, documentation, and utility functions
• Updated query definitions in test fixtures and conftest files
• Modified documentation examples to use the new word set reference

Reviewed Changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/test_utils.py	Updated query definitions to use unpleasant_5a
tests/test_query.py	Updated test cases and query objects to reference unpleasant_5a
tests/test_preprocessing.py	Updated query fixtures and test functions to use unpleasant_5a
tests/test_datasets.py	Updated expected word set list to include unpleasant_5a
tests/metrics/conftest.py	Updated query fixtures to use unpleasant_5a
tests/debias/test_multiclass_hard_debias.py	Updated attribute list to use unpleasant_5a
tests/debias/test_hard_debias.py	Updated attribute list to use unpleasant_5a
tests/debias/test_half_sibling_regression.py	Updated attribute list to use unpleasant_5a
tests/debias/test_double_hard_debias.py	Updated commented code to use unpleasant_5a
tests/debias/conftest.py	Updated query fixtures and labels to use unpleasant_5a
docs/user_guide/mitigation_user_guide.rst	Updated documentation examples to use unpleasant_5a
docs/user_guide/measurement_user_guide.rst	Updated documentation examples to use unpleasant_5a