Add SemanticKITTI LiDAR segmentation tutorial

[tejasstanley]

This PR adds a SemanticKITTI LiDAR segmentation tutorial for evaluating MMDetection3D models with PerceptionMetrics.

Changes

• Adds a notebook workflow for loading SemanticKITTI data and exporting labels into PerceptionMetrics train-ID format.
• Runs MMDetection3D MinkUNet inference through the existing mmdet3d model utility.
• Adds an MMDetection3D model-output ontology for the 0..18 SemanticKITTI class convention.
• Adds ontology translation from PerceptionMetrics train IDs to MMDetection3D output IDs during evaluation.
• Adds BEV visualization for predictions and ground truth with aligned class IDs.
• Adds notebook guidance explaining setup, data preparation, model loading, ontology translation, evaluation, and visualization.
• Adds links for SemanticKITTI data, MMDetection3D, and the model checkpoint.