[FSDP][optim_state_dict][5/N] Remove optim_inputs for sharded state_dict. (#89981)
The argument, `optim_inputs`, is being deprecated. Sharded optimizer state_dict APIs are not be used. It is safe to remove them.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/89981
Approved by: https://github.com/awgu
