* make audio requirements optional
* add changelog for version v0.2
* add numpy constraint for compatibility with tensorflow1.15
* update faq
* fix nlp requiring tensorflow
* add torchvision to multimodal dependency
* bump version from 0.2.1 to 0.2.2
* add warning msg when tensorflow is not installed
Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/9268278
1. refine quick start and pipeline doc
2. remove tf pytorch easynlp from requirements
3. lazy import for torch and tensorflow
4. test successfully on linux and mac intel cpu
5. update api doc
Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/8882373