Add explainers for nearest neighbor models

[Zaphoood]

Motivation and Context

This PR adds three new explainers for efficiently explaining sklearns KNearestNeighbor and RadiusNeighborClassifier models.

---

Public API Changes

• No Public API changes
• Yes, Public API changes (Details below)

---

How Has This Been Tested?

Unittests have been added that compare the results of the explainers' efficient algorithms to baseline values which are obtained by computing the exact Shapley Values of a ground-truth Game.

---

Checklist

• The changes have been tested locally.
• Documentation has been updated (if the public API or usage changes).
• An entry has been added to CHANGELOG.md (if relevant for users).
• The code follows the project's style guidelines.
• I have considered the impact of these changes on the public API.

---

[codecov]

Codecov Report

❌ Patch coverage is 95.11677% with 23 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/shapiq/explainer/nn/weighted_knn.py	91.72%	11 Missing ⚠️
src/shapiq/explainer/nn/games/wknn.py	92.72%	4 Missing ⚠️
src/shapiq/explainer/nn/base.py	96.22%	2 Missing ⚠️
src/shapiq/explainer/nn/games/base.py	95.45%	2 Missing ⚠️
src/shapiq/explainer/utils.py	81.81%	2 Missing ⚠️
src/shapiq/explainer/nn/games/_util.py	93.75%	1 Missing ⚠️
src/shapiq/explainer/nn/threshold_nn.py	97.95%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[Copilot]

Pull request overview

This PR introduces dedicated explainers for scikit-learn nearest-neighbor classifiers to support efficient data-valuation-style explanations, and adds accompanying benchmark games, tests, and documentation.

Changes:

• Add NN explainers (KNNExplainer, WeightedKNNExplainer, ThresholdNNExplainer) and benchmark games for exact/ground-truth comparisons.
• Extend explainer dispatch to recognize scikit-learn KNN/RadiusNeighbors classifiers and add utility APIs for converting first-order arrays to/from InteractionValues.
• Add unit tests/fixtures and a documentation notebook covering NN data valuation.

Reviewed changes

Copilot reviewed 27 out of 28 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
src/shapiq/explainer/utils.py	Adds explainer type dispatch for sklearn KNN/RadiusNeighbors models.
src/shapiq/explainer/nn/base.py	Introduces shared NN explainer base (data extraction + validation).
src/shapiq/explainer/nn/knn.py	Adds unweighted KNN explainer implementation.
src/shapiq/explainer/nn/weighted_knn.py	Adds weighted KNN explainer implementation.
src/shapiq/explainer/nn/threshold_nn.py	Adds radius/threshold NN explainer implementation.
src/shapiq/explainer/nn/_util.py	Adds shared NN helper validation + ignored-parameter warning.
src/shapiq/explainer/nn/games/*	Adds benchmark “Game” implementations for exact Shapley ground truth.
src/shapiq/interaction_values.py	Adds from_first_order_array / to_first_order_array helpers.
tests/shapiq/fixtures/data.py	Adds a small multi-class dataset fixture + feature constant changes.
tests/shapiq/fixtures/models.py	Adds sklearn NN model fixtures.
tests/shapiq/tests_unit/tests_explainer/tests_nn_explainers/*	Adds unit tests for NN explainers, games, and utilities.
tests/shapiq/tests_unit/test_interaction_values.py	Adds tests for first-order array conversion helpers.
docs/source/notebooks/tabular_notebooks/knn_data_valuation.ipynb	New notebook demonstrating NN explainers for data valuation.
docs/source/references.bib	Adds citations for Jia 2019 / Wang 2023 / Wang 2024.
CHANGELOG.md	Adds changelog entry for NN explainers.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[mmschlk]

I applied some minor changes I noticed going through the review. The biggest change is that I introduced a ValidNNExplainerIndices = Literal["SV"] that captures the meaning of the current KNN Explainers better (aka. only SV possible).

There is only one outstanding change that need to be resolved before we can merge this in: We need to remove all notebooks from the repo, as I want to merge this PR before I am merging the KNN PR in. This means all notebooks need to be converted into this sphinx gallery format and examples then live under examples/<folder/plot_.py. Notably, each example needs to be very quick and take not more than a couple of seconds. We will also have a way to have longer examples somewhere but currently I would like to emphasize speed here (which KNN explainers anyway are)