Adds trivial TreeExplainer computation and fixes a Bug in parsing xgboost models.

[mmschlk]

Generated with Copilot:

This pull request includes multiple changes to the shapiq project, focusing on bug fixes, feature enhancements, and improvements to the test fixtures. The most important changes include a bug fix for xgboost, a new prediction method, handling trivial computations for single-feature trees, and updates to the test fixtures.

Bug Fixes and Enhancements:

• Fixed a bug in TreeExplainer for xgboost where trees missing some features caused failures.
• Added a trivial computation for TreeSHAP-IQ when trees use only one feature, improving efficiency for such cases. [1] [2] [3] [4] [5]

New Methods:

• Introduced predict_one method in TreeModel to predict the output for a single instance.

Test Fixtures:

• Added support for xgboost models in the validation and model fixtures. [1] [2]
• Updated data.py fixtures to ensure deep copies of data are used, preventing unintended modifications during tests. [1] [2] [3]
• Refactored model fixtures to use the new dataset fixtures and ensure consistency across tests. [1] [2] [3] [4]