[$] Weighted interleaving for memory tiering

The kernel has, for many years, had the ability to control how memoryallocation is performed in systems with multiple NUMA nodes. Morerecently, NUMA nodes have also been pressed into service to representdifferent classes of memory; those nodes are now organized into tiersaccording to their performance characteristics. While memory-allocationpolicies can control the placement of pages at the NUMA-node level, thekernel provides no way to connect those policies with memory tiers. Thispatch series from Gregory Price aims to change this situation byallowing allocations to be placed across tiers in a weighted manner.