Enhance update usage from orphans

Currently, the _update_usage_from_orphans in the periodic tasks _update_available_resource just do a little work, getting the memory. It doesn't include the disk, numa_topology etc, which can cause the inaccurate resources reporting. For example, if the orphans instance use the huge pages, it will cause the hugepage to be inaccurate in the numa topology; If the orphans instance use the numa topology, it will cause the whole numa reported is inaccurate.