Fix: error on instance from batch_x missing from batch_y in nearest()

[Vuenc]

Fixed the error described in issue #163
I implemented the lenient approach from my reply there (error only on if some batch instance is present in batch_x, but missing from batch_y). I implemented this as-is for the CPU version, and an equivalent check on the ptr_x/ptr_y for the GPU version to catch the case where one of batch_x/batch_y is None.

I also checked if any of the other methods suffer from the same problem, but that's not the case. Only knn and radius take batch_x,batch_y arguments, and both seem to correctly handle empty instances (don't return any edge within an empty instance).

Additionally I implemented a check in nearest() if the batch_x, batch_y are sorted (as it's a general requirement of the method, and I also rely on it when calling torch.unique_consecutive()).

[codecov-commenter]

Codecov Report

Merging #168 (4c3ce21) into master (82e9df9) will increase coverage by 0.05%.
The diff coverage is 100.00%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

@@ Coverage Diff @@ ## master #168 +/- ## ========================================== + Coverage 98.83% 98.88% +0.05%  ========================================== Files 10 10 Lines 86 90 +4 ========================================== + Hits 85 89 +4  Misses 1 1 

Impacted Files	Coverage Δ
torch_cluster/nearest.py	100.00% <100.00%> (ø)

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

[rusty1s]

Thank you!