BLD: add autoawq in setup

[utopia2077]

Support autoawq.
Otherwise, when inferring the AWQ quantization model, it will error that dependencies are missing.

[qinxuye]

I think you can add the autoawq to ALL and transformers.

Refer to auto-gptq

inference/setup.cfg

Line 97 in 2f1ee95

auto-gptq ; sys_platform!='darwin'

and

inference/setup.cfg

Line 117 in 2f1ee95

auto-gptq

Then when pip install "xinference[all]" it will be installed, so you don't need to modify the dockerfile.

[utopia2077]

Modified as above.
Installing dependencies inside a Docker container is very inconvenient, so it is recommended to add it.

[qinxuye]

LGTM