gfpgan still downloading some files on first usage

[ssube]

Some part of the GFPGAN code path is still downloading PTH files rather than using the cached ones:

Downloading: "https://github.com/xinntao/facexlib/releases/download/v0.1.0/detection_Resnet50_Final.pth" to /home/ssube/onnx-web/api/gfpgan/weights/detection_Resnet50_Final.pth
                                                                                                                                                                                                      
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 104M/104M [00:02<00:00, 42.6MB/s]
Downloading: "https://github.com/xinntao/facexlib/releases/download/v0.2.2/parsing_parsenet.pth" to /home/ssube/onnx-web/api/gfpgan/weights/parsing_parsenet.pth
                                                                                                                                                                                                      
 11%|█████████████████▊                                                                                                                                          | 9.29M/81.4M [00:00<00:02, 36.1MB/s]
[2023-02-14 00:22:18,782] INFO: werkzeug: 10.2.2.16 - - [14/Feb/2023 00:22:18] "GET /api/ready?output=txt2img_148066626_98ae8818a6b1ad4bacc13bd174a63bef525e0b733a29a2aa0e34b836138df968_1676334097.pn
g HTTP/1.1" 200 -                                                                                  
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 81.4M/81.4M [00:02<00:00, 41.2MB/s]
[2023-02-14 00:22:20,873] DEBUG: onnx_web.server.model_cache: Adding new model to cache: gfpgan                                                                                                       
[2023-02-14 00:22:20,874] INFO: onnx_web.server.model_cache: Removing models from cache, 1 of 4