Fix make_model_config_json function

[thanawan-atc]

Description

Fix make_model_config_json function so that we can trace the model without having to create the model config file by ourselves.

Have been tested on tracing the following models in TorchScript and ONNX format:
sentence-transformers/msmarco-distilbert-base-tas-b
sentence-transformers/multi-qa-MiniLM-L6-cos-v1
sentence-transformers/all-MiniLM-L12-v2
sentence-transformers/multi-qa-mpnet-base-dot-v1

Issues Resolved

• Allow user to specify the model_format
• Enable it to scrape pooling_mode from Huggingface
• Enable it to scrape normalize_result from Huggingface
• Fix a bug in scraping embedding_dimension

Check List

• Commits are signed per the DCO using --signoff

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[codecov]

Codecov Report

Merging #188 (dcb0755) into main (976a512) will increase coverage by 0.15%.
The diff coverage is 92.30%.

@@            Coverage Diff             @@
##             main     #188      +/-   ##
==========================================
+ Coverage   90.81%   90.96%   +0.15%     
==========================================
  Files          37       37              
  Lines        4050     4073      +23     
==========================================
+ Hits         3678     3705      +27     
+ Misses        372      368       -4     

Impacted Files	Coverage Δ
...search_py_ml/ml_models/sentencetransformermodel.py	70.55% <92.30%> (+2.90%)	⬆️

[dhrubo-os]

Thanks for adding all those tests.