add ability to chain together multiple hotstart runs for long wall-clock times

[ghost]

such that if the user specifies a wall clock time that is longer than the platform's job limit time, then it is split into multiple hotstart runs. The successive hotstart runs would be very similar to the first, except that instead of linking to the spinup run and being Slurm dependent upon the spinup run they would instead link to the previous hotstart run and be Slurm dependent on it