Implement a pull-through VirtualMemory Cache

[ko1N]

Due to the nature of physical memory access it is sometimes inconvenient that data is paged out and inaccessible because it did not change in a while. A pull-through VirtualMemory Cache (similar to the PhysicalMemory caches) could mitigate this issue or at least reduce it a bit. Ideally the cache size is fully configurable.

It is probably best to start working on this issue after the unification of vmem/pmem.
Additionally the "user" needs to be able to access exactly which individual page lookups failed during bigger reads.