The striped allocator divides the address space in one private
range and a number of chunks. The idea is that each process/machine
is assigned one chunk, and can allocate into other processes
into its own chunk at predictable addresses. This will be used
for O(1) importing of CNF.
This depends on D1264.