* add self-hosted repo:
* add extra requirements for different field and reduce necessary requirements
* update docker file with so required by audio
* add requirements checker which will be used later when implement lazy import
* remove repeated requirements and replace opencv-python-headless with opencv-python

example usage:
```shell
pip install model_scope[all] -f https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/release/maas/repo.html
pip install model_scope[cv] -f https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/release/maas/repo.html
pip install model_scope[nlp] -f https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/release/maas/repo.html
pip install model_scope[audio] -f https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/release/maas/repo.html
pip install model_scope[multi-modal] -f https://pai-vision-data-hz.oss-cn-zhangjiakou.aliyuncs.com/release/maas/repo.html

```
Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/9211383

* format pipeline output and check it
* fix UT
* add docstr to clarify the difference between model.postprocess and pipeline.postprocess

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/9051405

* add dockerfile
* uninstall opencv-python-headless
* update develop doc
Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/9020302