Added CPU support

[zolastro]

Fix #9

CUDA was always used when using torch model, even if CPU was selected.

Changes:

• Use self._ctxs and self._devices instead of overriding this variables.
• Check if using GPU or CPU before calling DataParallel, since it always uses GPU.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.