fix float max_neighbors passed to nvalchemi

[joehart2001]

Summary

Fix a type error in the CUDA/nvalchemi neighbour-list backend for large cutoffs.

#259 added a max_neighbors estimate:

len(system) * max(128, cutoff**3)

cutoff is a float, so this expression is also a float once cutoff**3 > 128 (i.e. above roughly 5 Å). However, nvalchemi requires max_neighbors to be an integer.

I noticed this whilst running benchmarks on ml-peg with pet-cam-xl. this can be reproduced by running e.g. the lattice constant benchmark on gpu.

Contributor (creator of pull-request) checklist

• Tests updated (for new features and bugfixes)?
• Documentation updated (for new features)?
• Issue referenced (for PRs that solve an issue)?

Reviewer checklist

• CHANGELOG updated with public API or any other important changes?

[frostedoyster]

Thanks!