[HPU] Enable torch.jit.load for HPU (#81759)
As per torch.jit.load documentation, all previously saved modules,
irrespective of their device, are first loaded onto CPU, and then
are moved to the devices they were saved from. So far, supported
devices included CPU and CUDA only. To enable torch.jit.load for
HPU, additional check for HPU is introduced.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/81759
Approved by: https://github.com/eellison
