Refactor particle prev_data and vault cleanup

[folex]

Currently, vm::cleanup call (here), is implemented wrong:

if let Some(vm) = self.vm_pool.get_vm() {
  self.actors.retain(|actor| 
    if actos.is_expired { vm.cleanup(actor.particle_id) }
  )
  ...

The problem here is that actor clean is dependent on VM availability which doesn't make any sense: we don't need WASM runtime to call a bunch of std::fs::remove.

What should be done instead, is remove cleanup and initialize from aquavm::DataStore and manage this solely on the node.