generate alpha (4 channel) based output in swapping and enhancing

Hi,

How do you generate 4 channel (RGBA) based output image in swapping and enhancing?
Here is my code:
```python
source = cv2.imread('d:/source.jpg')
target = cv2.imread('d:/target.jpg')

source_face = FACE_ANALYSER.get(source)[0]
target_face = FACE_ANALYSER.get(target)[0]

swapped = FACE_SWAPPER.get(target, target_face, source_face, paste_back=True)
enhanced = enhance_face(target_face, result)

cv2.imwrite('d:/out.jpg', enhanced)
```
I need enhanced/swapped face in (RGBA) format, which later can be composited on target.jpg with more editing.
The [enhance_face](https://github.com/s0md3v/roop/blob/0497d0ba029ba24881c8c4d8644189909dc0def8/roop/processors/frame/face_enhancer.py#L61) function.