This was neccessary in order to make the memory available via GDRcopy
when multiple small allocations were made. cudaMalloc() would return
multiple memory chunks located in the same GPU page, which GDRcopy
pretty much dislikes (`gdrdrv:offset != 0 is not supported`).
As a side effect, this will keep the number of BAR-mappings done
via GDRcopy low, because they seem to be quite limited.

This is useful when we sub-delegate management of a memory block
to another allocator.