Implement Popular Link Prediction Models

[HarryShomer]

🚀 The feature, motivation and pitch

Link prediction (LP) is a fundamental graph task. However, PyG currently lacks implementations of many newer methods designed specifically for link prediction. I think many people would find it useful to have easy to use implementations of LP methods integrated into PyG.

I compiled an (inexhaustive) list of methods that I think are worth implementing.

1. NCN/NCNC
2. BUDDY
3. SEAL - An example is included in examples/seal_link_pred.py
4. LPFormer - Open PR at Add LPFormer model and example #9956
5. Neo-GNN
6. MPLP/MPLP+ - Open PR at Add MPLP link prediction model and example #9420

I think adding these models in nn.models would be a great addition to PyG.

If anyone is willing to contribute, please feel free to chime in! My current plan is to slowly work my way through them.

Alternatives

No response

Additional context

No response