UCT/CUDA_IPC: Implemented uct_cuda_ipc_rkey_ptr

[Yiltan]

What

Implements the function uct_cuda_ipc_rkey_ptr.

Why ?

So that we can call ucp_rkey_ptr on CUDA memory.

This would allow us to write to a remote processes GPU's memory within a GPU kernel.

__global__ void device_function(double *addr_p, double *A, double *B, double *C) 
{
    int i = get_thread_ID();
    // Do work
    addr_p[i] = A[i] * B[i] + C[i];
}

__host__ void host_function(...)
{
    // Do work
    status = ucp_rkey_ptr(rkey, raddr, &addr_p);
    // Do work
    device_function<<<...>>>(addr_p, A, B, C);
}

[Yiltan]

@Akshay-Venkatesh @bureddy

I was wondering if this addition to the cuda_ipc module could be up for discussion?