[Model] Add CoED-GNN model with multi-dataset node classification support

[K1llery]

Description

Add the CoED-GNN (Continuous Edge Direction Graph Neural Network) model from the paper "Improving Graph Neural
Networks by Learning Continuous Edge Directions" to GammaGL.

The implementation includes a directional convolution layer based on fuzzy Laplacian, a backbone model with
jumping-knowledge support, and a standard GammaGL training flow for node classification on 5 benchmark datasets.

Checklist

• The PR title starts with [Model]
• Changes are complete
• All changes have test coverage
• Code is well-documented
• To the best of my knowledge, examples are either not affected by this change, or have been fixed to be
  compatible with this change
• Related issue is referred in this PR

Changes

• gammagl/layers/conv/coed_conv.py: CoED directional convolution layer inheriting MessagePassing, with
  separate src-to-dst and dst-to-src message passing and optional self-feature transform
• gammagl/models/coed.py: CoEDModel backbone with multi-layer CoED convolution, jumping-knowledge aggregation
  (cat/max/lstm), dropout, and L2 normalization
• examples/coed/coed_trainer.py: Standard GammaGL training flow using SemiSpvzLoss + TrainOneStep,
  supporting 5 datasets (Cora, Texas, Wisconsin, Chameleon, Squirrel) with Geom-GCN 10 fixed splits
• examples/coed/readme.md: Documentation with paper link, dataset statistics, and run commands

Supported Datasets

Dataset	# Nodes	# Edges	# Classes
Cora	2,708	10,556	7
Texas	183	309	5
Wisconsin	251	515	5
Chameleon	2,277	36,101	5
Squirrel	5,201	217,073	5

Verified Results (CPU, torch backend)

Dataset	Original coed-gnn	GammaGL
Texas	84.05 ± 4.09	82.97 ± 4.02
Wisconsin	86.86 ± 5.34	84.51 ± 4.51

Run Commands

TL_BACKEND="torch" python examples/coed/coed_trainer.py --dataset cora
TL_BACKEND="torch" python examples/coed/coed_trainer.py --dataset texas
TL_BACKEND="torch" python examples/coed/coed_trainer.py --dataset wisconsin
TL_BACKEND="torch" python examples/coed/coed_trainer.py --dataset chameleon
TL_BACKEND="torch" python examples/coed/coed_trainer.py --dataset squirrel