[BUG]: Ensure cudaMemcpy is called by thrust::copy

We should ensure that this:

```c++
void test(double* in, double* out, size_t n) {
thrust::copy(thrust::device, in, in + n, out);
}
```

calls `cudaMemcpy` or `cudaMemcpyAsync` with `cudaMemcpyDefault`.

Right now it does not seem to be happening. @jrhemstad

Start with the thrust::copy implementation and its existing tests; reproduce the shown device-to-device case and trace which CUDA copy operation is used. Done means a regression test verifies that the operation uses cudaMemcpy or cudaMemcpyAsync with cudaMemcpyDefault.

Written by the indexing model from the issue text.