Add a GATConv model using the primitives in cugraph-ops

[tingyu66]

This PR adds a GAT model torch_geometric.nn.GATConvCuGraph that leverages the optimized aggregation function in cugraph-ops, similar to #6278.

Fixes https://github.com/rapidsai/cugraph-ops/issues/182.

[codecov]

Codecov Report

Merging #6412 (18c9c64) into master (ab3efed) will decrease coverage by 1.61%.
The diff coverage is 92.30%.

❗ Current head 18c9c64 differs from pull request most recent head 45afb9b. Consider uploading reports for the commit 45afb9b to get more accurate results

@@            Coverage Diff             @@
##           master    #6412      +/-   ##
==========================================
- Coverage   87.85%   86.25%   -1.61%     
==========================================
  Files         416      417       +1     
  Lines       22808    22779      -29     
==========================================
- Hits        20039    19649     -390     
- Misses       2769     3130     +361     

Impacted Files	Coverage Δ
torch_geometric/nn/conv/cugraph/sage_conv.py	100.00% <ø> (+20.00%)	⬆️
torch_geometric/nn/conv/cugraph/base.py	85.71% <75.00%> (-14.29%)	⬇️
torch_geometric/nn/conv/__init__.py	100.00% <100.00%> (ø)
torch_geometric/nn/conv/cugraph/__init__.py	100.00% <100.00%> (ø)
torch_geometric/nn/conv/cugraph/gat_conv.py	100.00% <100.00%> (ø)
torch_geometric/nn/models/dimenet_utils.py	0.00% <0.00%> (-75.52%)	⬇️
torch_geometric/nn/models/dimenet.py	17.84% <0.00%> (-49.66%)	⬇️
torch_geometric/visualization/graph.py	68.25% <0.00%> (-26.99%)	⬇️
torch_geometric/nn/conv/utils/typing.py	83.75% <0.00%> (-15.00%)	⬇️
torch_geometric/nn/pool/asap.py	92.00% <0.00%> (-8.00%)	⬇️
... and 38 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[tingyu66]

@rusty1s GATConv updated. I removed add_self_loops option here, since it is better for users to change the COO directly IMO.