Fix of cusolver error on GPU

[Ghassen-Chaabouni]

Description

A fix for this error

torch\nn\functional.py:718: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at  ..\c10/core/TensorImpl.h:1156.)
  return torch.max_pool2d(input, kernel_size, stride, padding, dilation, ceil_mode)
torch\nn\functional.py:3609: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details.
  warnings.warn(
Exception in Tkinter callback

Traceback (most recent call last):
  File "tkinter\__init__.py", line 1892, in __call__
  File "customtkinter\windows\widgets\ctk_button.py", line 554, in _clicked
    self._command()
  File "ui.py", line 702, in start_button_event
  File "dot\__main__.py", line 67, in run
  File "dot\dot.py", line 131, in generate
  File "dot\commons\model_option.py", line 185, in generate_from_camera
  File "dot\commons\camera_utils.py", line 71, in camera_pipeline
  File "dot\simswap\option.py", line 180, in process_image
  File "dot\simswap\util\reverse2original.py", line 136, in reverse2wholeimage
RuntimeError: cusolver error: CUSOLVER_STATUS_INTERNAL_ERROR, when calling `cusolverDnCreate(handle)`

Using torch.linalg.inv results in an error on GPU RTX 4000 series.

Fixes #94, #106

Changelog:

Added:

• Added cupy to the requirements

Updated:

• everse2original.py to use cp.linalg.inv instead of torch.linalg.inv on GPU

Fixed:

• cusolver error on GPU

[Ghassen-Chaabouni]

I generated the executable. It's available here. It needs to be tested.

[Ghassen-Chaabouni]

This error is due the torch.linalg.inv function. I replaced it with the same function from cupy.
The executable was tested and this change seemed to fix the error
#106 (comment)