support int4 weight only quant for llama3

[FlamingoPg]

Support int4 weight only quantization for Llama3

1. Define the weight only layer in ModelParallel.py
2. Define ConvertWeightToOpmx.py and add quant here
3. Update Dynamic Static modeling for quant model
4. Fix the Woqu function for packed model
5. Add weight only quant README.md
6. Test export and demo under static and dynamic batching on A100 GPU

Some WIP:

1. Support AutoAWQ when ConvertWeightToOpmx

[FlamingoPg]

@Jzz24 hi, I finish this PR. Can you review it?

[Jzz24]

@Jzz24 hi, I finish this PR. Can you review it?

OK

[FlamingoPg]

用naive 量化实现的Llama3模型导出，后续补上awq的。讨论后决定暂不做SplitModel 和 MergeModel

[FlamingoPg]

上个PR没合，我就在那个基础上继续写了。代码太多可能review起来有些困难QAQ

整体的逻辑是这样的，我把量化的步骤定义在了ConvertWeightToOpmx.py里，这个文件完成量化的工作。之后在Modeling里我适配了量化。量化层的定义也写在了ModelParallel.py里。