[atomate2] restart for long running HSE jobs

I’m running some HSE jobs and they tend to hit wall time.
When I restart the calculations with:
lpad rerun_fws -s RUNNING --task-level -pd
The calculation will go into the previous directory but VASP will use the POSCAR to restart thus wasting all the calculation time used to obtain the CONTCAR.

I think we should build a mechanism in atomate2 to use the CONTCAR for restarting.