[FEA]: Replace cuco::detail::MurmurHash3_32 with a proper public API

[PointKernel]

Is this a new feature, an improvement, or a change to existing functionality?

Improvement

How would you describe the priority of this feature request

High

Please provide a clear description of problem this feature solves

NVIDIA/cuCollections#310 introduced a breaking change that the content of cuco/detail/hash_functions.cuh has been reorganized to a different place. As a result, many header inclusions in cugraph will crash, e.g.:

cugraph/cpp/tests/prims/mg_transform_reduce_v.cu

Line 32 in afd79ec

#include <cuco/detail/hash_functions.cuh>

This will be an issue if we want to bump the cuco version in rapids-cmake.

Describe your ideal solution

Include the public hasher header and choose a proper hash function in cugraph, e.g.:

#include <cuco/hash_functions.cuh>

...

void func() {
  auto hasher = cuco::murmurhash3_32<vertex_t>{};
}

Note cuco provides new 32-bit hasher options like cuco::xxhash_32 which is proved to be more efficient than murmurhash. Worth testing in cugraph.

Describe any alternatives you have considered

Use cuco default hasher: https://github.com/NVIDIA/cuCollections/blob/806aa8051ba933c758636586e2c34487282465f0/include/cuco/hash_functions.cuh#L76

Additional context

No response

Code of Conduct

• I agree to follow cuGraph's Code of Conduct
• I have searched the open feature requests and have found no duplicates for this feature request