feat: initial public release v0.1.0 — MLX port of pyannote-speaker-diarization-3.1

Byte-parity with pyannote-PyTorch reference (cosine 0.763718 identical
at 6 decimals on 200 cross-window slot pairs). 2.5x faster than
pyannote-MPS on Apple Silicon native.

Extracted from gitea.tavportal.com/olivier/MLX_CONVERTOR commit 5f9eafa.

tests/unit/test_diar_segmentation_load.py

@@ -0,0 +1,9 @@
+"""Unit test: load SegmentationModel weights from HF mlx-community repo."""
+import pytest
+from pyannote_diarization_3_1_mlx.segmentation import SegmentationModel
+
+
+@pytest.mark.integration
+def test_segmentation_loads_from_hf():
+    m = SegmentationModel.from_hf()
+    assert m is not None