Merge Graph-EFM model from `prob_model_lam` branch

[joeloskarsson]

This issue is to keep track of merging of the prob_model_lam branch into main. This branch contains the Graph-EFM model from Oskarsson et al. (2024) for LAM.

The main changes in this branch include:

• [ ] Addition of PropagationNet GNN layers https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/interaction_net.py#L134 Make these optionally usable also in existing deterministic models.
• [ ] Add new metric: Sample-based CRPS https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/metrics.py#L235
• [ ] Add new metric: Squared ensemble spread https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/metrics.py#L328 Remove square to make just ensemble spread before logging.
• [ ] Add the Graph-EFM model itself https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/models/graph_efm.py#L17 This includes the different sub-modules (latent encoders and decoders).
• [ ] (Potentially) rename the Hi-LAM model to Graph-FM https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/models/graph_fm.py#L9
• [ ] Add plotting functions for ensemble forecasts https://github.com/mllam/neural-lam/blob/a12bceb6ca3e021a3f896247ece150ee49c7bd37/neural_lam/vis.py#L116

I think it is best to do this merge (or at least the addition of the Graph-EFM model) after #49 is done, so the new model can fit into the new class hierarchy.