[OMNIML-2244] enable fp8 and int8 ONNX export

[ajrasane]

What does this PR do?

Type of change:
Example update

Overview:

• Support ONNX export for fp8 and int8 precisions
• Added utility functions to check for fp8 and int8 quantization (will be used in ONNXExporter)
• Fixed a bug in evaluation API for high batch sizes
• Added function to replace zeros from scales to smallest positive value in fp16

Usage

python torch_quant_to_onnx.py \
    --quantize_mode fp8/int8 \
    --onnx_save_path <onnx_path>  

Testing

Validated the accuracy and latency of int8 and fp8 models:

Metric	INT8	FP8	FP16
Top1 Accuracy	84.584%	85.062%	85.102%
Top5 Accuracy	97.3%	97.534%	97.526%

Before your PR is "Ready for review"

• Make sure you read and follow Contributor guidelines and your commits are signed.
• Is this change backward compatible?: Yes
• Did you write any new necessary tests?: No
• Did you add or update any necessary documentation?: Yes
• Did you update Changelog?: No

[codecov]

Codecov Report

❌ Patch coverage is 13.04348% with 20 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.57%. Comparing base (a5025a2) to head (fa88100).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
modelopt/onnx/quantization/qdq_utils.py	8.33%	11 Missing ⚠️
modelopt/torch/_deploy/utils/torch_onnx.py	18.18%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #594      +/-   ##
==========================================
- Coverage   74.64%   74.57%   -0.08%     
==========================================
  Files         183      183              
  Lines       18389    18412      +23     
==========================================
+ Hits        13727    13730       +3     
- Misses       4662     4682      +20     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[galagam]

LGTM

[cjluo-nv]

could you document in what condition do we need to call this method here?

[ajrasane]

Sorry, I had already set the MR to auto merge. Will add this in a follow up MR.