out of ~1000000 invocations:
  rnd.cc generates 999426 odd and 510 even numbers
  arc4.hh generates 500412 odd and 499508 even numbers

( make HW=user o.user/utest && ./o.user/utest )

no need for vmap::lock anymore either

than modifying the vma's type and vmnode inplace.

about what fields can/cannot be changed by another core
while the object is allocated.

minor fix up: mv hwvm.hh include/hwvm.hh

Conflicts:
	qlock.h

more flexible updatepages() interface
PTE locking (using one of the free PTE bits)
guard against concurrent unmaps (except no way to check if vma is deleted yet..)