Update build_tree function with SparseKmeans implementation

[khoinpd0411]

What does this PR do?

Update build_tree function with SparseKmeans implementation and utilize an adaptive clustering method mixing Elkan's and Lloyd's algorithm based on the number of samples.

Improvements:

• Speed up tree construction.
• Resolved convergence issues caused by duplicate samples during clustering
• Introduced an adaptive clustering strategy that dynamically switches between Elkan’s algorithm (for large sample sizes) and Lloyd’s algorithm (for smaller or dense datasets)

Test CLI & API (bash tests/autotest.sh)

Test APIs used by main.py.

• Test Pass
  • (Copy and paste the last outputted line here.)
• Not Applicable (i.e., the PR does not include API changes.)

Check API Document

If any new APIs are added, please check if the description of the APIs is added to API document.

• API document is updated (linear, nn)
• Not Applicable (i.e., the PR does not include API changes.)

Test quickstart & API (bash tests/docs/test_changed_document.sh)

If any APIs in quickstarts or tutorials are modified, please run this test to check if the current examples can run correctly after the modified APIs are released.

[Eleven1Liu]

Please also add sparsekmeans to install_requires

LibMultiLabel/setup.cfg

Lines 27 to 35 in a0bef91

	install_requires =
	liblinear-multicore>=2.49.0
	numba
	pandas>1.3.0
	PyYAML
	scikit-learn
	scipy<1.14.0
	tqdm
	psutil

[Eleven1Liu]

Please bump version to 0.8.0

LibMultiLabel/setup.cfg

Line 3 in a0bef91

version = 0.7.4

[zhi-bao]

Sparsekmeans requires Python >= 3.10, whereas LibMultiLabel supports Python >= 3.8.
This causes installation issues when users try to install LibMultiLabel in Python 3.8 or 3.9.
There are two approach for this issue:

• Update LibMultiLabel to require Python >= 3.10.
• Or, detect the user's environment and apply the corresponding workaround.
  There is much room for discussion.

[Eleven1Liu]

@khoinpd0411 No need to bump version now, we will release with #20.

[chcwww]

I'm just wondering why the indentation isn't aligned with line:287.
BTW, should we pass the verbose flag through _build_tree() so that we can control the output when training?
And would it be better to do something like (I'm not sure)

if label_representation.shape[0] > 10000:
    kmeans_algo = ElkanKmeans
else:
    kmeans_algo = LloydKmeans
kmeans = kmeans_algo(those_params)

[Eleven1Liu]

For the formatting issue, please use black formatter.

[maclin726]

Maybe it's better to have

num_unique_labels = len(np.unique(metalabels))
if len(num_unique_labels) == K:
    children = []
    for i in range(K):
        child_representation = label_representation[metalabels == i]
        child_map = label_map[metalabels == i]
        child = _build_tree(child_representation, child_map, d + 1, K, dmax)
        children.append(child)
else:
    children = [
        Node(label_map=label_map[metalabels == i], children=[])
            for i in range(num_unique_labels)
    ]