HypergraphConv TorchScript Compilation Compatibility

[chrisrosner]

Addresses #10399

This was the minimal set of changes I could figure out to get HypergraphConv to compile with torchscript.

Included are unit tests demonstrating the problems.

The following issues were resolved:
type of alpha changing from None -> tensor
torchscript not understanding the import functional as F
non-attention branch does not define some member variables
default value for dropout is wrong type

Test Plan

pytest -k hyper

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.94%. Comparing base (c211214) to head (a6bf19f).
⚠️ Report is 94 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #10400      +/-   ##
==========================================
- Coverage   86.11%   85.94%   -0.18%     
==========================================
  Files         496      502       +6     
  Lines       33655    35130    +1475     
==========================================
+ Hits        28981    30191    +1210     
- Misses       4674     4939     +265     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wsad1]

@chrisrosner thanks for the work.
Could you enable Maintainers are allowed to edit this pull request. for this PR. Need to merge master so that tests pass.

[wsad1]

Suggested change

	output = script(torch.randn(4, in_channels),
	torch.tensor([[0, 1, 2], [0, 0, 1]]),
	hyperedge_attr=torch.randn(2, in_channels))
	assert output.size() == (4, out_channels * conv.heads)
	x = torch.randn(4, in_channels)
	out_1 = conv(x, torch.tensor([[0, 1, 2], [0, 0, 1]]),
	hyperedge_attr=torch.randn(2, in_channels))
	out_2 = script(x,
	torch.tensor([[0, 1, 2], [0, 0, 1]]),
	hyperedge_attr=torch.randn(2, in_channels))
	assert torch.allclose(out_1, out_2)

Also add the onlyFullTest decorator like

pytorch_geometric/test/nn/pool/test_pool.py

Line 10 in b2a1fdc

@onlyFullTest

.
Good to merge after this.