Use clamp with bincount approach

[kmontemayor2-sc]

This got brought up in #273 (comment) -- I was unaware of clamp (or that it worked like this!) so let's fix this elsewhere :)

This fixes an issue where we could potentially see nodes get removed if their count mod 255 = 0, but clamp fixes this, see:

>>> torch.tensor(255, dtype=torch.uint8).add_(1).clamp(255)
tensor(255, dtype=torch.uint8)
>>> torch.tensor(255, dtype=torch.uint8).add_(1)
tensor(0, dtype=torch.uint8)

Thanks @svij-sc !

[svij-sc]

Thanks.

[kmontemayor2-sc]

/unit_test

[kmontemayor2-sc]

/integration_test

[kmontemayor2-sc]

/test

[github-actions]

GiGL Automation

@ 19:08:12UTC : 🔄 Unit Test started.

@ 19:47:08UTC : ✅ Workflow completed successfully.

[github-actions]

GiGL Automation

@ 19:08:18UTC : 🔄 Integration Test started.

@ 20:03:45UTC : ✅ Workflow completed successfully.