Add image segmentation support to GUI

[tejasstanley]

This PR extends the GUI structure to support image segmentation alongside the existing image detection flow.
Changes include:

• Adds task-specific image segmentation sidebar, dataset viewer, inference, and evaluator tabs.
• Adds Cityscapes support for the image segmentation GUI.
• Adds shared GUI utilities under tabs/tasks/utils.py.
• Moves browse helpers out of perceptionmetrics/utils/gui.py.
• Uses TorchImageSegmentationModel for segmentation inference and evaluation.
• Adds live evaluation progress callbacks for segmentation evaluation.

[dpascualhe]

Image segmentation is not a placeholder anymore

[dpascualhe]

There are a couple of regressions here due to some recent changes in object detection pre/post-processing. Namely, the min_value in the max. detections widget, the option for Max Side in the resizing strategy and the logic for retrieving padding info (you can check it out and compare against https://github.com/JdeRobot/PerceptionMetrics/blob/master/app.py)