Multiple model's supported by VSGAN

[rlaphoenix]

I'm thinking VSGAN's next venture for perhaps v2.0 is to support different models.
Perhaps I could even add support for the user to specify a prepared model to use!

Right now VSGAN only supports ESRGAN (Enhanced-SRGAN). There are now new SRGAN enhancements, or rather ESRGAN Enhancements like ESRGAN+ and Real-ESRGAN now too, perhaps VSGAN could one day support these.

Models that might be nice to support:

• SwinIR; Image Restoration Using Swin Transformer
• GFPGAN; Practical Algorithm for Real-world Face Restoration

[rlaphoenix]

ESRGAN+ and Real-ESRGAN (including 2x and 1x) are now supported as of 0387807