feat: support large ONNX model files

[ZiniuYu]

NOTICE Please merge #825 first!!!

This PR supports the following ONNX models:

• ViT-H-14::laion2B-s32B-b79K
• ViT-g-14::laion2B-s12B-b42K
• M-CLIP/XLM-Roberta-Large-Vit-B-32
• M-CLIP/XLM-Roberta-Large-Vit-L-14
• M-CLIP/XLM-Roberta-Large-Vit-B-16Plus
• M-CLIP/LABSE-Vit-L-14

This is accomplished by rewriting the loading logic: if the model file is a zip file, we unzip it to a tmp dir and load the model from there.

This PR also rewrites loading M-CLIP logic: the visual model now loads from our custom open-clip models

[codecov]

Codecov Report

Merging #828 (4b754e3) into main (09d1548) will increase coverage by 1.80%.
The diff coverage is 52.94%.

@@            Coverage Diff             @@
##             main     #828      +/-   ##
==========================================
+ Coverage   82.57%   84.38%   +1.80%     
==========================================
  Files          21       21              
  Lines        1567     1575       +8     
==========================================
+ Hits         1294     1329      +35     
+ Misses        273      246      -27     

Flag	Coverage Δ
cas	84.38% <52.94%> (+1.80%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
server/clip_server/model/clip_onnx.py	78.18% <41.66%> (-10.71%)	⬇️
server/clip_server/model/mclip_model.py	83.33% <80.00%> (-0.88%)	⬇️
server/clip_server/model/clip_model.py	91.66% <0.00%> (+4.16%)	⬆️
server/clip_server/model/clip_trt.py	85.71% <0.00%> (+16.32%)	⬆️
server/clip_server/model/trt_utils.py	83.51% <0.00%> (+27.47%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[github-actions]

📝 Docs are deployed on https://ft-zip-model--jina-docs.netlify.app 🎉