Add functions for loading spatial nodes and visualization

[Steriva]

This pull request adds new visualization capabilities for spatial data, particularly for datasets with 2D spatial structure, and introduces a utility for loading mesh/node information. The main changes are the addition of spatial average and 2D contour plotting functions, along with support for overlaying a "blanket" region for the MSFR dataset.

Visualization enhancements:

• Added plot_spatial_averages and plot_2d_contours methods to the Visualizer class in visualization_module.py to enable plotting of spatial averages over time and 2D contour comparisons between truth and predictions. This includes logic for handling multiple fields and predictions, as well as flexible configuration options.
• Updated the plot_from_batch method to support new plot types: "spatial_averages" and "contour2d", including loading mesh nodes and handling dataset-specific overlays.
• Added add_blanket_msfr function to overlay a white rectangle representing the blanket region in the MSFR dataset on 2D contour plots.

Data loading utility:

• Introduced a new load_nodes function in data_module.py to load spatial mesh/node data (nodes.npy) for a given dataset, raising an error if the file is missing.

[yyexela]

Looks good to me! Ran naive baseline and verified all plots generated and looked ok. Thanks Stefano!