Switch RMM device_memory_resource to device_async_resource_ref in Java code

[pmattione-nvidia]

This replaces the RMM device_memory_resource with device_async_resource_ref in the Java code. This also updates the memory resources to conform to the new cuda::memory_resource async allocator concept.

Note that the C++ RMM memory resources that are ultimately created by the initialize() function still inherit from device_memory_resource. Thus this work cannot be completed until RMM is updated to provide a way to create equivalent resources that follow the new cuda::memory_resource async allocator concepts.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.